iSCSI Gateway reboots and permanent loss
We utilize 4 iSCSI gateways in a cluster and have noticed the following during patching cycles when we sequentially reboot single iSCSI-gateways: "gwcli" often hangs on the still-up iSCSI GWs but sometimes still functions and gives the message: "1 gateway is inaccessible - updates will be disabled" This got me thinking about what the course of action would be should an iSCSI gateway fail permanently or semi-permanently, say a hardware issue. What would be the best course of action to instruct the remaining iSCSI gateways that one of them is no longer available and that they should allow updates again and take ownership of the now-defunct-node's LUNS? I'm guessing pulling down the RADOS config object and rewriting it and re-put'ing it followed by a rbd-target-api restart might do the trick but am hoping there is a more "in-band" and less potentially devastating way to do this. Thanks for any insights. Respectfully, *Wes Dillingham* wes@wesdillingham.com LinkedIn <http://www.linkedin.com/in/wesleydillingham>
If I recall correctly, the recent ceph-iscsi release supports the removal of a gateway via the "gwcli". I think the Ceph dashboard can do that as well. On Tue, Dec 3, 2019 at 1:59 PM Wesley Dillingham <wes@wesdillingham.com> wrote:
We utilize 4 iSCSI gateways in a cluster and have noticed the following during patching cycles when we sequentially reboot single iSCSI-gateways:
"gwcli" often hangs on the still-up iSCSI GWs but sometimes still functions and gives the message:
"1 gateway is inaccessible - updates will be disabled"
This got me thinking about what the course of action would be should an iSCSI gateway fail permanently or semi-permanently, say a hardware issue. What would be the best course of action to instruct the remaining iSCSI gateways that one of them is no longer available and that they should allow updates again and take ownership of the now-defunct-node's LUNS?
I'm guessing pulling down the RADOS config object and rewriting it and re-put'ing it followed by a rbd-target-api restart might do the trick but am hoping there is a more "in-band" and less potentially devastating way to do this.
Thanks for any insights.
Respectfully,
Wes Dillingham wes@wesdillingham.com LinkedIn _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io
-- Jason
Gateway removal is indeed supported since ceph-iscsi 3.0 (or was it 2.7?) and it works while it is offline :) Paul -- Paul Emmerich Looking for help with your Ceph cluster? Contact us at https://croit.io croit GmbH Freseniusstr. 31h 81247 München www.croit.io Tel: +49 89 1896585 90 On Tue, Dec 3, 2019 at 8:21 PM Jason Dillaman <jdillama@redhat.com> wrote:
If I recall correctly, the recent ceph-iscsi release supports the removal of a gateway via the "gwcli". I think the Ceph dashboard can do that as well.
On Tue, Dec 3, 2019 at 1:59 PM Wesley Dillingham <wes@wesdillingham.com> wrote:
We utilize 4 iSCSI gateways in a cluster and have noticed the following during patching cycles when we sequentially reboot single iSCSI-gateways:
"gwcli" often hangs on the still-up iSCSI GWs but sometimes still functions and gives the message:
"1 gateway is inaccessible - updates will be disabled"
This got me thinking about what the course of action would be should an iSCSI gateway fail permanently or semi-permanently, say a hardware issue. What would be the best course of action to instruct the remaining iSCSI gateways that one of them is no longer available and that they should allow updates again and take ownership of the now-defunct-node's LUNS?
I'm guessing pulling down the RADOS config object and rewriting it and re-put'ing it followed by a rbd-target-api restart might do the trick but am hoping there is a more "in-band" and less potentially devastating way to do this.
Thanks for any insights.
Respectfully,
Wes Dillingham wes@wesdillingham.com LinkedIn _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io
-- Jason _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io
I do not think it's going to do what you want when the node you want to delete is down. It looks like we only temporarily stop the gw from being exported. It does not update the gateway.cfg, because we do the config removal call on the node we want to delete. So gwcli would report success and the ls command will show it as no longer running/exported, but if you restart the rbd-target-api service then it will show up again. There is an internal command to do what you want. I will post a PR for gwlci and so it can be used by dashboard. On 12/03/2019 01:19 PM, Jason Dillaman wrote:
If I recall correctly, the recent ceph-iscsi release supports the removal of a gateway via the "gwcli". I think the Ceph dashboard can do that as well.
On Tue, Dec 3, 2019 at 1:59 PM Wesley Dillingham <wes@wesdillingham.com> wrote:
We utilize 4 iSCSI gateways in a cluster and have noticed the following during patching cycles when we sequentially reboot single iSCSI-gateways:
"gwcli" often hangs on the still-up iSCSI GWs but sometimes still functions and gives the message:
"1 gateway is inaccessible - updates will be disabled"
This got me thinking about what the course of action would be should an iSCSI gateway fail permanently or semi-permanently, say a hardware issue. What would be the best course of action to instruct the remaining iSCSI gateways that one of them is no longer available and that they should allow updates again and take ownership of the now-defunct-node's LUNS?
I'm guessing pulling down the RADOS config object and rewriting it and re-put'ing it followed by a rbd-target-api restart might do the trick but am hoping there is a more "in-band" and less potentially devastating way to do this.
Thanks for any insights.
Respectfully,
Wes Dillingham wes@wesdillingham.com LinkedIn _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io
Thanks. If I am reading this correctly the ability to remove an iSCSI gateway would allow the remaining iSCSI gateways to take over for the removed gateway's LUN's as of > 3.0. Thats good, we run 3.2. However, because the actual update of the central config object happens from the to-be-deleted iSCSI gateway, despite where the gwcli command is issued, it will fail to actually remove said gateway from the object if that gateway is not functioning. I guess this leaves the question still of how to proceed when one of the iSCSI gateways fails permanently? Is that possible, or is it potentially possible other than manually intervening on the config object? If its not possible would the best course of action be to have standby hardware and quickly recreate the node or perhaps run the gateways more ephemerally, from a VM or container? Thanks again. Respectfully, *Wes Dillingham* wes@wesdillingham.com LinkedIn <http://www.linkedin.com/in/wesleydillingham> On Tue, Dec 3, 2019 at 2:45 PM Mike Christie <mchristi@redhat.com> wrote:
I do not think it's going to do what you want when the node you want to delete is down.
It looks like we only temporarily stop the gw from being exported. It does not update the gateway.cfg, because we do the config removal call on the node we want to delete.
So gwcli would report success and the ls command will show it as no longer running/exported, but if you restart the rbd-target-api service then it will show up again.
There is an internal command to do what you want. I will post a PR for gwlci and so it can be used by dashboard.
On 12/03/2019 01:19 PM, Jason Dillaman wrote:
If I recall correctly, the recent ceph-iscsi release supports the removal of a gateway via the "gwcli". I think the Ceph dashboard can do that as well.
On Tue, Dec 3, 2019 at 1:59 PM Wesley Dillingham <wes@wesdillingham.com> wrote:
We utilize 4 iSCSI gateways in a cluster and have noticed the following
during patching cycles when we sequentially reboot single iSCSI-gateways:
"gwcli" often hangs on the still-up iSCSI GWs but sometimes still
functions and gives the message:
"1 gateway is inaccessible - updates will be disabled"
This got me thinking about what the course of action would be should an
iSCSI gateway fail permanently or semi-permanently, say a hardware issue. What would be the best course of action to instruct the remaining iSCSI gateways that one of them is no longer available and that they should allow updates again and take ownership of the now-defunct-node's LUNS?
I'm guessing pulling down the RADOS config object and rewriting it and
re-put'ing it followed by a rbd-target-api restart might do the trick but am hoping there is a more "in-band" and less potentially devastating way to do this.
Thanks for any insights.
Respectfully,
Wes Dillingham wes@wesdillingham.com LinkedIn _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io
On 12/03/2019 04:19 PM, Wesley Dillingham wrote:
Thanks. If I am reading this correctly the ability to remove an iSCSI gateway would allow the remaining iSCSI gateways to take over for the removed gateway's LUN's as of > 3.0. Thats good, we run 3.2. However, because the actual update of the central config object happens from the to-be-deleted iSCSI gateway, despite where the gwcli command is issued, it will fail to actually remove said gateway from the object if that gateway is not functioning.
Yes.
I guess this leaves the question still of how to proceed when one of the iSCSI gateways fails permanently? Is that possible, or is it potentially possible other than manually intervening on the config
You could edit the gateway.cfg manually, but I would not do it, because it's error prone. It's probably safest to run in degraded mode and wait for an updated ceph-iscsi package with a fix. If you are running into the problem right now, I can bump the priority.
object? If its not possible would the best course of action be to have standby hardware and quickly recreate the node or perhaps run the gateways more ephemerally, from a VM or container?
Thanks again.
Respectfully,
*Wes Dillingham* wes@wesdillingham.com <mailto:wes@wesdillingham.com> LinkedIn <http://www.linkedin.com/in/wesleydillingham>
On Tue, Dec 3, 2019 at 2:45 PM Mike Christie <mchristi@redhat.com <mailto:mchristi@redhat.com>> wrote:
I do not think it's going to do what you want when the node you want to delete is down.
It looks like we only temporarily stop the gw from being exported. It does not update the gateway.cfg, because we do the config removal call on the node we want to delete.
So gwcli would report success and the ls command will show it as no longer running/exported, but if you restart the rbd-target-api service then it will show up again.
There is an internal command to do what you want. I will post a PR for gwlci and so it can be used by dashboard.
On 12/03/2019 01:19 PM, Jason Dillaman wrote: > If I recall correctly, the recent ceph-iscsi release supports the > removal of a gateway via the "gwcli". I think the Ceph dashboard can > do that as well. > > On Tue, Dec 3, 2019 at 1:59 PM Wesley Dillingham <wes@wesdillingham.com <mailto:wes@wesdillingham.com>> wrote: >> >> We utilize 4 iSCSI gateways in a cluster and have noticed the following during patching cycles when we sequentially reboot single iSCSI-gateways: >> >> "gwcli" often hangs on the still-up iSCSI GWs but sometimes still functions and gives the message: >> >> "1 gateway is inaccessible - updates will be disabled" >> >> This got me thinking about what the course of action would be should an iSCSI gateway fail permanently or semi-permanently, say a hardware issue. What would be the best course of action to instruct the remaining iSCSI gateways that one of them is no longer available and that they should allow updates again and take ownership of the now-defunct-node's LUNS? >> >> I'm guessing pulling down the RADOS config object and rewriting it and re-put'ing it followed by a rbd-target-api restart might do the trick but am hoping there is a more "in-band" and less potentially devastating way to do this. >> >> Thanks for any insights. >> >> Respectfully, >> >> Wes Dillingham >> wes@wesdillingham.com <mailto:wes@wesdillingham.com> >> LinkedIn >> _______________________________________________ >> ceph-users mailing list -- ceph-users@ceph.io <mailto:ceph-users@ceph.io> >> To unsubscribe send an email to ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io> > > >
Hi, Em qua., 4 de dez. de 2019 às 00:31, Mike Christie <mchristi@redhat.com> escreveu:
On 12/03/2019 04:19 PM, Wesley Dillingham wrote:
Thanks. If I am reading this correctly the ability to remove an iSCSI gateway would allow the remaining iSCSI gateways to take over for the removed gateway's LUN's as of > 3.0. Thats good, we run 3.2. However, because the actual update of the central config object happens from the to-be-deleted iSCSI gateway, despite where the gwcli command is issued, it will fail to actually remove said gateway from the object if that gateway is not functioning.
Yes.
I guess this leaves the question still of how to proceed when one of the iSCSI gateways fails permanently? Is that possible, or is it potentially possible other than manually intervening on the config
You could edit the gateway.cfg manually, but I would not do it, because it's error prone.
It's probably safest to run in degraded mode and wait for an updated ceph-iscsi package with a fix. If you are running into the problem right now, I can bump the priority.
I permanently lost a gateway. I can not leave running "degraded" because I need to add another redundancy gateway, and it does not allow with the gateway "offline".
In this case, what can I do? If I create a new gateway with the same name and IP as the lost one, and then try to use "delete" in gwcli, will it work?
object? If its not possible would the best course of action be to have standby hardware and quickly recreate the node or perhaps run the gateways more ephemerally, from a VM or container?
Thanks again.
Respectfully,
*Wes Dillingham* wes@wesdillingham.com <mailto:wes@wesdillingham.com> LinkedIn <http://www.linkedin.com/in/wesleydillingham>
On Tue, Dec 3, 2019 at 2:45 PM Mike Christie <mchristi@redhat.com <mailto:mchristi@redhat.com>> wrote:
I do not think it's going to do what you want when the node you want to delete is down.
It looks like we only temporarily stop the gw from being exported. It does not update the gateway.cfg, because we do the config removal call on the node we want to delete.
So gwcli would report success and the ls command will show it as no longer running/exported, but if you restart the rbd-target-api service then it will show up again.
There is an internal command to do what you want. I will post a PR for gwlci and so it can be used by dashboard.
On 12/03/2019 01:19 PM, Jason Dillaman wrote: > If I recall correctly, the recent ceph-iscsi release supports the > removal of a gateway via the "gwcli". I think the Ceph dashboard can > do that as well. > > On Tue, Dec 3, 2019 at 1:59 PM Wesley Dillingham <wes@wesdillingham.com <mailto:wes@wesdillingham.com>> wrote: >> >> We utilize 4 iSCSI gateways in a cluster and have noticed the following during patching cycles when we sequentially reboot single iSCSI-gateways: >> >> "gwcli" often hangs on the still-up iSCSI GWs but sometimes still functions and gives the message: >> >> "1 gateway is inaccessible - updates will be disabled" >> >> This got me thinking about what the course of action would be should an iSCSI gateway fail permanently or semi-permanently, say a hardware issue. What would be the best course of action to instruct the remaining iSCSI gateways that one of them is no longer available and that they should allow updates again and take ownership of the now-defunct-node's LUNS? >> >> I'm guessing pulling down the RADOS config object and rewriting it and re-put'ing it followed by a rbd-target-api restart might do the trick but am hoping there is a more "in-band" and less potentially devastating way to do this. >> >> Thanks for any insights. >> >> Respectfully, >> >> Wes Dillingham >> wes@wesdillingham.com <mailto:wes@wesdillingham.com> >> LinkedIn >> _______________________________________________ >> ceph-users mailing list -- ceph-users@ceph.io <mailto:ceph-users@ceph.io> >> To unsubscribe send an email to ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io> > > >
_______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io
On Wed, Dec 4, 2019 at 9:27 AM Gesiel Galvão Bernardes <gesiel.bernardes@gmail.com> wrote:
Hi,
Em qua., 4 de dez. de 2019 às 00:31, Mike Christie <mchristi@redhat.com> escreveu:
On 12/03/2019 04:19 PM, Wesley Dillingham wrote:
Thanks. If I am reading this correctly the ability to remove an iSCSI gateway would allow the remaining iSCSI gateways to take over for the removed gateway's LUN's as of > 3.0. Thats good, we run 3.2. However, because the actual update of the central config object happens from the to-be-deleted iSCSI gateway, despite where the gwcli command is issued, it will fail to actually remove said gateway from the object if that gateway is not functioning.
Yes.
I guess this leaves the question still of how to proceed when one of the iSCSI gateways fails permanently? Is that possible, or is it potentially possible other than manually intervening on the config
You could edit the gateway.cfg manually, but I would not do it, because it's error prone.
It's probably safest to run in degraded mode and wait for an updated ceph-iscsi package with a fix. If you are running into the problem right now, I can bump the priority.
I permanently lost a gateway. I can not leave running "degraded" because I need to add another redundancy gateway, and it does not allow with the gateway "offline".
In this case, what can I do? If I create a new gateway with the same name and IP as the lost one, and then try to use "delete" in gwcli, will it work?
Yes, it will work if they are configured w/ the same name + IPs and you re-install the ceph-iscsi SW.
object? If its not possible would the best course of action be to have standby hardware and quickly recreate the node or perhaps run the gateways more ephemerally, from a VM or container?
Thanks again.
Respectfully,
*Wes Dillingham* wes@wesdillingham.com <mailto:wes@wesdillingham.com> LinkedIn <http://www.linkedin.com/in/wesleydillingham>
On Tue, Dec 3, 2019 at 2:45 PM Mike Christie <mchristi@redhat.com <mailto:mchristi@redhat.com>> wrote:
I do not think it's going to do what you want when the node you want to delete is down.
It looks like we only temporarily stop the gw from being exported. It does not update the gateway.cfg, because we do the config removal call on the node we want to delete.
So gwcli would report success and the ls command will show it as no longer running/exported, but if you restart the rbd-target-api service then it will show up again.
There is an internal command to do what you want. I will post a PR for gwlci and so it can be used by dashboard.
On 12/03/2019 01:19 PM, Jason Dillaman wrote: > If I recall correctly, the recent ceph-iscsi release supports the > removal of a gateway via the "gwcli". I think the Ceph dashboard can > do that as well. > > On Tue, Dec 3, 2019 at 1:59 PM Wesley Dillingham <wes@wesdillingham.com <mailto:wes@wesdillingham.com>> wrote: >> >> We utilize 4 iSCSI gateways in a cluster and have noticed the following during patching cycles when we sequentially reboot single iSCSI-gateways: >> >> "gwcli" often hangs on the still-up iSCSI GWs but sometimes still functions and gives the message: >> >> "1 gateway is inaccessible - updates will be disabled" >> >> This got me thinking about what the course of action would be should an iSCSI gateway fail permanently or semi-permanently, say a hardware issue. What would be the best course of action to instruct the remaining iSCSI gateways that one of them is no longer available and that they should allow updates again and take ownership of the now-defunct-node's LUNS? >> >> I'm guessing pulling down the RADOS config object and rewriting it and re-put'ing it followed by a rbd-target-api restart might do the trick but am hoping there is a more "in-band" and less potentially devastating way to do this. >> >> Thanks for any insights. >> >> Respectfully, >> >> Wes Dillingham >> wes@wesdillingham.com <mailto:wes@wesdillingham.com> >> LinkedIn >> _______________________________________________ >> ceph-users mailing list -- ceph-users@ceph.io <mailto:ceph-users@ceph.io> >> To unsubscribe send an email to ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io> > > >
_______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io
_______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io
-- Jason
On 12/04/2019 08:26 AM, Gesiel Galvão Bernardes wrote:
Hi,
Em qua., 4 de dez. de 2019 às 00:31, Mike Christie <mchristi@redhat.com <mailto:mchristi@redhat.com>> escreveu:
On 12/03/2019 04:19 PM, Wesley Dillingham wrote: > Thanks. If I am reading this correctly the ability to remove an iSCSI > gateway would allow the remaining iSCSI gateways to take over for the > removed gateway's LUN's as of > 3.0. Thats good, we run 3.2. However, > because the actual update of the central config object happens from the > to-be-deleted iSCSI gateway, despite where the gwcli command is issued, > it will fail to actually remove said gateway from the object if that > gateway is not functioning.
Yes.
> > I guess this leaves the question still of how to proceed when one of the > iSCSI gateways fails permanently? Is that possible, or is it > potentially possible other than manually intervening on the config
You could edit the gateway.cfg manually, but I would not do it, because it's error prone.
It's probably safest to run in degraded mode and wait for an updated ceph-iscsi package with a fix. If you are running into the problem right now, I can bump the priority.
I permanently lost a gateway. I can not leave running "degraded" because I need to add another redundancy gateway, and it does not allow with the gateway "offline".
In this case, what can I do? If I create a new gateway with the same name and IP as the lost one, and then try to use "delete" in gwcli, will it work?
Yes. If you can have a temp stop in services you can also do the following as a workaround: 0. Stop applications accessing iscsi luns, and have the initiator log out of the iscsi target. 1. Stop ceph iscsi service. On all iscsi gw nodes do: systemctl stop rbd-target-api 2. Delete gateway.cfg. This will delete the configuration info like the target and its ACL and LUN mappings. It does not delete the actual images or pools that you have data on. rados -p rbd rm gateway.cfg 3. Start ceph iscsi services again. On all iscsi gw nodes do: systemctl start rbd-target-api 4. Resetup target with gwcli. For the image/disk setup stage, instead of doing the "create" command do the "attach"command: attach pool=your_pool image=image_name Then just re-add your target, ACLs and LUN mappings. 5. On the initiator side relogin to the iscsi target.
> object? If its not possible would the best course of action be to have > standby hardware and quickly recreate the node or perhaps run the > gateways more ephemerally, from a VM or container? > > Thanks again. > > Respectfully, > > *Wes Dillingham* > wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>> > LinkedIn <http://www.linkedin.com/in/wesleydillingham> > > > On Tue, Dec 3, 2019 at 2:45 PM Mike Christie <mchristi@redhat.com <mailto:mchristi@redhat.com> > <mailto:mchristi@redhat.com <mailto:mchristi@redhat.com>>> wrote: > > I do not think it's going to do what you want when the node you want to > delete is down. > > It looks like we only temporarily stop the gw from being exported. It > does not update the gateway.cfg, because we do the config removal call > on the node we want to delete. > > So gwcli would report success and the ls command will show it as no > longer running/exported, but if you restart the rbd-target-api service > then it will show up again. > > There is an internal command to do what you want. I will post a PR for > gwlci and so it can be used by dashboard. > > > On 12/03/2019 01:19 PM, Jason Dillaman wrote: > > If I recall correctly, the recent ceph-iscsi release supports the > > removal of a gateway via the "gwcli". I think the Ceph dashboard can > > do that as well. > > > > On Tue, Dec 3, 2019 at 1:59 PM Wesley Dillingham > <wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>>> wrote: > >> > >> We utilize 4 iSCSI gateways in a cluster and have noticed the > following during patching cycles when we sequentially reboot single > iSCSI-gateways: > >> > >> "gwcli" often hangs on the still-up iSCSI GWs but sometimes still > functions and gives the message: > >> > >> "1 gateway is inaccessible - updates will be disabled" > >> > >> This got me thinking about what the course of action would be > should an iSCSI gateway fail permanently or semi-permanently, say a > hardware issue. What would be the best course of action to instruct > the remaining iSCSI gateways that one of them is no longer available > and that they should allow updates again and take ownership of the > now-defunct-node's LUNS? > >> > >> I'm guessing pulling down the RADOS config object and rewriting > it and re-put'ing it followed by a rbd-target-api restart might do > the trick but am hoping there is a more "in-band" and less > potentially devastating way to do this. > >> > >> Thanks for any insights. > >> > >> Respectfully, > >> > >> Wes Dillingham > >> wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>> > >> LinkedIn > >> _______________________________________________ > >> ceph-users mailing list -- ceph-users@ceph.io <mailto:ceph-users@ceph.io> > <mailto:ceph-users@ceph.io <mailto:ceph-users@ceph.io>> > >> To unsubscribe send an email to ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io> > <mailto:ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io>> > > > > > > > _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io <mailto:ceph-users@ceph.io> To unsubscribe send an email to ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io>
I have never had a permanent loss of a gateway but I'm a believer in Murphy's law and want to have a plan. Glad to hear that there is a solution in-the-works, curious when might that be available in a release? If sooner than later I'll plan to upgrade then immediately, otherwise, if far down the queue I would like to know if I should ready a standby server. Thanks so much for all your great work on this product. Respectfully, *Wes Dillingham* wes@wesdillingham.com LinkedIn <http://www.linkedin.com/in/wesleydillingham> On Wed, Dec 4, 2019 at 11:18 AM Mike Christie <mchristi@redhat.com> wrote:
Hi,
Em qua., 4 de dez. de 2019 às 00:31, Mike Christie <mchristi@redhat.com <mailto:mchristi@redhat.com>> escreveu:
On 12/03/2019 04:19 PM, Wesley Dillingham wrote: > Thanks. If I am reading this correctly the ability to remove an iSCSI > gateway would allow the remaining iSCSI gateways to take over for
On 12/04/2019 08:26 AM, Gesiel Galvão Bernardes wrote: the
> removed gateway's LUN's as of > 3.0. Thats good, we run 3.2.
However,
> because the actual update of the central config object happens from the > to-be-deleted iSCSI gateway, despite where the gwcli command is issued, > it will fail to actually remove said gateway from the object if
that
> gateway is not functioning.
Yes.
> > I guess this leaves the question still of how to proceed when one of the > iSCSI gateways fails permanently? Is that possible, or is it > potentially possible other than manually intervening on the config
You could edit the gateway.cfg manually, but I would not do it,
because
it's error prone.
It's probably safest to run in degraded mode and wait for an updated ceph-iscsi package with a fix. If you are running into the problem
right
now, I can bump the priority.
I permanently lost a gateway. I can not leave running "degraded" because I need to add another redundancy gateway, and it does not allow with the gateway "offline".
In this case, what can I do? If I create a new gateway with the same name and IP as the lost one, and then try to use "delete" in gwcli, will it work?
Yes.
If you can have a temp stop in services you can also do the following as a workaround:
0. Stop applications accessing iscsi luns, and have the initiator log out of the iscsi target.
1. Stop ceph iscsi service. On all iscsi gw nodes do:
systemctl stop rbd-target-api
2. Delete gateway.cfg. This will delete the configuration info like the target and its ACL and LUN mappings. It does not delete the actual images or pools that you have data on.
rados -p rbd rm gateway.cfg
3. Start ceph iscsi services again. On all iscsi gw nodes do:
systemctl start rbd-target-api
4. Resetup target with gwcli. For the image/disk setup stage, instead of doing the "create" command do the "attach"command:
attach pool=your_pool image=image_name
Then just re-add your target, ACLs and LUN mappings.
5. On the initiator side relogin to the iscsi target.
> object? If its not possible would the best course of action be to
have
> standby hardware and quickly recreate the node or perhaps run the > gateways more ephemerally, from a VM or container? > > Thanks again. > > Respectfully, > > *Wes Dillingham* > wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>> > LinkedIn <http://www.linkedin.com/in/wesleydillingham> > > > On Tue, Dec 3, 2019 at 2:45 PM Mike Christie <mchristi@redhat.com <mailto:mchristi@redhat.com> > <mailto:mchristi@redhat.com <mailto:mchristi@redhat.com>>> wrote: > > I do not think it's going to do what you want when the node you want to > delete is down. > > It looks like we only temporarily stop the gw from being exported. It > does not update the gateway.cfg, because we do the config removal call > on the node we want to delete. > > So gwcli would report success and the ls command will show it as no > longer running/exported, but if you restart the rbd-target-api service > then it will show up again. > > There is an internal command to do what you want. I will post a PR for > gwlci and so it can be used by dashboard. > > > On 12/03/2019 01:19 PM, Jason Dillaman wrote: > > If I recall correctly, the recent ceph-iscsi release supports the > > removal of a gateway via the "gwcli". I think the Ceph dashboard can > > do that as well. > > > > On Tue, Dec 3, 2019 at 1:59 PM Wesley Dillingham > <wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>>>
wrote:
> >> > >> We utilize 4 iSCSI gateways in a cluster and have noticed
the
> following during patching cycles when we sequentially reboot single > iSCSI-gateways: > >> > >> "gwcli" often hangs on the still-up iSCSI GWs but sometimes still > functions and gives the message: > >> > >> "1 gateway is inaccessible - updates will be disabled" > >> > >> This got me thinking about what the course of action would
be
> should an iSCSI gateway fail permanently or semi-permanently, say a > hardware issue. What would be the best course of action to instruct > the remaining iSCSI gateways that one of them is no longer available > and that they should allow updates again and take ownership of
the
> now-defunct-node's LUNS? > >> > >> I'm guessing pulling down the RADOS config object and
rewriting
> it and re-put'ing it followed by a rbd-target-api restart
might do
> the trick but am hoping there is a more "in-band" and less > potentially devastating way to do this. > >> > >> Thanks for any insights. > >> > >> Respectfully, > >> > >> Wes Dillingham > >> wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>> > >> LinkedIn > >> _______________________________________________ > >> ceph-users mailing list -- ceph-users@ceph.io <mailto:ceph-users@ceph.io> > <mailto:ceph-users@ceph.io <mailto:ceph-users@ceph.io>> > >> To unsubscribe send an email to ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io> > <mailto:ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io>> > > > > > > > _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io <mailto:ceph-users@ceph.io> To unsubscribe send an email to ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io>
_______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io
On 12/04/2019 02:34 PM, Wesley Dillingham wrote:
I have never had a permanent loss of a gateway but I'm a believer in Murphy's law and want to have a plan. Glad to hear that there is a solution in-the-works, curious when might that be available in a release? If sooner than later I'll plan to upgrade then immediately,
It should be in the next release which I think we would just make when the patch gets merged since we have a good number of fixes sitting in the repo. Patch/PR is here https://github.com/ceph/ceph-iscsi/pull/156 if you have a non production setup and are used to applying patches and testing upstream.
otherwise, if far down the queue I would like to know if I should ready a standby server.
Thanks so much for all your great work on this product.
Respectfully,
*Wes Dillingham* wes@wesdillingham.com <mailto:wes@wesdillingham.com> LinkedIn <http://www.linkedin.com/in/wesleydillingham>
On Wed, Dec 4, 2019 at 11:18 AM Mike Christie <mchristi@redhat.com <mailto:mchristi@redhat.com>> wrote:
On 12/04/2019 08:26 AM, Gesiel Galvão Bernardes wrote: > Hi, > > Em qua., 4 de dez. de 2019 às 00:31, Mike Christie <mchristi@redhat.com <mailto:mchristi@redhat.com> > <mailto:mchristi@redhat.com <mailto:mchristi@redhat.com>>> escreveu: > > On 12/03/2019 04:19 PM, Wesley Dillingham wrote: > > Thanks. If I am reading this correctly the ability to remove an iSCSI > > gateway would allow the remaining iSCSI gateways to take over for the > > removed gateway's LUN's as of > 3.0. Thats good, we run 3.2. However, > > because the actual update of the central config object happens > from the > > to-be-deleted iSCSI gateway, despite where the gwcli command is > issued, > > it will fail to actually remove said gateway from the object if that > > gateway is not functioning. > > Yes. > > > > > I guess this leaves the question still of how to proceed when one > of the > > iSCSI gateways fails permanently? Is that possible, or is it > > potentially possible other than manually intervening on the config > > You could edit the gateway.cfg manually, but I would not do it, because > it's error prone. > > It's probably safest to run in degraded mode and wait for an updated > ceph-iscsi package with a fix. If you are running into the problem right > now, I can bump the priority. > > I permanently lost a gateway. I can not leave running "degraded" because > I need to add another redundancy gateway, and it does not allow with the > gateway "offline". > > In this case, what can I do? If I create a new gateway with the same > name and IP as the lost one, and then try to use "delete" in gwcli, will > it work?
Yes.
If you can have a temp stop in services you can also do the following as a workaround:
0. Stop applications accessing iscsi luns, and have the initiator log out of the iscsi target.
1. Stop ceph iscsi service. On all iscsi gw nodes do:
systemctl stop rbd-target-api
2. Delete gateway.cfg. This will delete the configuration info like the target and its ACL and LUN mappings. It does not delete the actual images or pools that you have data on.
rados -p rbd rm gateway.cfg
3. Start ceph iscsi services again. On all iscsi gw nodes do:
systemctl start rbd-target-api
4. Resetup target with gwcli. For the image/disk setup stage, instead of doing the "create" command do the "attach"command:
attach pool=your_pool image=image_name
Then just re-add your target, ACLs and LUN mappings.
5. On the initiator side relogin to the iscsi target.
> > > > > object? If its not possible would the best course of action be to have > > standby hardware and quickly recreate the node or perhaps run the > > gateways more ephemerally, from a VM or container? > > > > Thanks again. > > > > Respectfully, > > > > *Wes Dillingham* > > wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>> > <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>>> > > LinkedIn <http://www.linkedin.com/in/wesleydillingham> > > > > > > On Tue, Dec 3, 2019 at 2:45 PM Mike Christie <mchristi@redhat.com <mailto:mchristi@redhat.com> > <mailto:mchristi@redhat.com <mailto:mchristi@redhat.com>> > > <mailto:mchristi@redhat.com <mailto:mchristi@redhat.com> <mailto:mchristi@redhat.com <mailto:mchristi@redhat.com>>>> wrote: > > > > I do not think it's going to do what you want when the node > you want to > > delete is down. > > > > It looks like we only temporarily stop the gw from being > exported. It > > does not update the gateway.cfg, because we do the config > removal call > > on the node we want to delete. > > > > So gwcli would report success and the ls command will show it > as no > > longer running/exported, but if you restart the rbd-target-api > service > > then it will show up again. > > > > There is an internal command to do what you want. I will post > a PR for > > gwlci and so it can be used by dashboard. > > > > > > On 12/03/2019 01:19 PM, Jason Dillaman wrote: > > > If I recall correctly, the recent ceph-iscsi release > supports the > > > removal of a gateway via the "gwcli". I think the Ceph > dashboard can > > > do that as well. > > > > > > On Tue, Dec 3, 2019 at 1:59 PM Wesley Dillingham > > <wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>> > <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>>>> wrote: > > >> > > >> We utilize 4 iSCSI gateways in a cluster and have noticed the > > following during patching cycles when we sequentially reboot > single > > iSCSI-gateways: > > >> > > >> "gwcli" often hangs on the still-up iSCSI GWs but sometimes > still > > functions and gives the message: > > >> > > >> "1 gateway is inaccessible - updates will be disabled" > > >> > > >> This got me thinking about what the course of action would be > > should an iSCSI gateway fail permanently or semi-permanently, > say a > > hardware issue. What would be the best course of action to > instruct > > the remaining iSCSI gateways that one of them is no longer > available > > and that they should allow updates again and take ownership of the > > now-defunct-node's LUNS? > > >> > > >> I'm guessing pulling down the RADOS config object and rewriting > > it and re-put'ing it followed by a rbd-target-api restart might do > > the trick but am hoping there is a more "in-band" and less > > potentially devastating way to do this. > > >> > > >> Thanks for any insights. > > >> > > >> Respectfully, > > >> > > >> Wes Dillingham > > >> wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>> > <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>>> > > >> LinkedIn > > >> _______________________________________________ > > >> ceph-users mailing list -- ceph-users@ceph.io <mailto:ceph-users@ceph.io> > <mailto:ceph-users@ceph.io <mailto:ceph-users@ceph.io>> > > <mailto:ceph-users@ceph.io <mailto:ceph-users@ceph.io> <mailto:ceph-users@ceph.io <mailto:ceph-users@ceph.io>>> > > >> To unsubscribe send an email to ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io> > <mailto:ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io>> > > <mailto:ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io> > <mailto:ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io>>> > > > > > > > > > > > > _______________________________________________ > ceph-users mailing list -- ceph-users@ceph.io <mailto:ceph-users@ceph.io> > <mailto:ceph-users@ceph.io <mailto:ceph-users@ceph.io>> > To unsubscribe send an email to ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io> > <mailto:ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io>> > _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io <mailto:ceph-users@ceph.io> To unsubscribe send an email to ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io>
Thats great thank you so much. I will try and get this patch in my test env asap but will likely wait for official release cut for prod. I really appreciate you adding this in to the product. Respectfully, *Wes Dillingham* wes@wesdillingham.com LinkedIn <http://www.linkedin.com/in/wesleydillingham> On Thu, Dec 5, 2019 at 4:14 PM Mike Christie <mchristi@redhat.com> wrote:
On 12/04/2019 02:34 PM, Wesley Dillingham wrote:
I have never had a permanent loss of a gateway but I'm a believer in Murphy's law and want to have a plan. Glad to hear that there is a solution in-the-works, curious when might that be available in a release? If sooner than later I'll plan to upgrade then immediately,
It should be in the next release which I think we would just make when the patch gets merged since we have a good number of fixes sitting in the repo.
Patch/PR is here
https://github.com/ceph/ceph-iscsi/pull/156
if you have a non production setup and are used to applying patches and testing upstream.
otherwise, if far down the queue I would like to know if I should ready a standby server.
Thanks so much for all your great work on this product.
Respectfully,
*Wes Dillingham* wes@wesdillingham.com <mailto:wes@wesdillingham.com> LinkedIn <http://www.linkedin.com/in/wesleydillingham>
On Wed, Dec 4, 2019 at 11:18 AM Mike Christie <mchristi@redhat.com <mailto:mchristi@redhat.com>> wrote:
On 12/04/2019 08:26 AM, Gesiel Galvão Bernardes wrote: > Hi, > > Em qua., 4 de dez. de 2019 às 00:31, Mike Christie <mchristi@redhat.com <mailto:mchristi@redhat.com> > <mailto:mchristi@redhat.com <mailto:mchristi@redhat.com>>> escreveu: > > On 12/03/2019 04:19 PM, Wesley Dillingham wrote: > > Thanks. If I am reading this correctly the ability to remove an iSCSI > > gateway would allow the remaining iSCSI gateways to take over for the > > removed gateway's LUN's as of > 3.0. Thats good, we run 3.2. However, > > because the actual update of the central config object happens > from the > > to-be-deleted iSCSI gateway, despite where the gwcli command is > issued, > > it will fail to actually remove said gateway from the object if that > > gateway is not functioning. > > Yes. > > > > > I guess this leaves the question still of how to proceed when one > of the > > iSCSI gateways fails permanently? Is that possible, or is it > > potentially possible other than manually intervening on the config > > You could edit the gateway.cfg manually, but I would not do it, because > it's error prone. > > It's probably safest to run in degraded mode and wait for an updated > ceph-iscsi package with a fix. If you are running into the problem right > now, I can bump the priority. > > I permanently lost a gateway. I can not leave running "degraded" because > I need to add another redundancy gateway, and it does not allow with the > gateway "offline". > > In this case, what can I do? If I create a new gateway with the same > name and IP as the lost one, and then try to use "delete" in gwcli, will > it work?
Yes.
If you can have a temp stop in services you can also do the following as a workaround:
0. Stop applications accessing iscsi luns, and have the initiator log out of the iscsi target.
1. Stop ceph iscsi service. On all iscsi gw nodes do:
systemctl stop rbd-target-api
2. Delete gateway.cfg. This will delete the configuration info like the target and its ACL and LUN mappings. It does not delete the actual images or pools that you have data on.
rados -p rbd rm gateway.cfg
3. Start ceph iscsi services again. On all iscsi gw nodes do:
systemctl start rbd-target-api
4. Resetup target with gwcli. For the image/disk setup stage, instead of doing the "create" command do the "attach"command:
attach pool=your_pool image=image_name
Then just re-add your target, ACLs and LUN mappings.
5. On the initiator side relogin to the iscsi target.
> > > > > object? If its not possible would the best course of action be to have > > standby hardware and quickly recreate the node or perhaps run the > > gateways more ephemerally, from a VM or container? > > > > Thanks again. > > > > Respectfully, > > > > *Wes Dillingham* > > wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>> > <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>>> > > LinkedIn <http://www.linkedin.com/in/wesleydillingham> > > > > > > On Tue, Dec 3, 2019 at 2:45 PM Mike Christie <mchristi@redhat.com <mailto:mchristi@redhat.com> > <mailto:mchristi@redhat.com <mailto:mchristi@redhat.com>> > > <mailto:mchristi@redhat.com <mailto:mchristi@redhat.com> <mailto:mchristi@redhat.com <mailto:mchristi@redhat.com>>>> wrote: > > > > I do not think it's going to do what you want when the node > you want to > > delete is down. > > > > It looks like we only temporarily stop the gw from being > exported. It > > does not update the gateway.cfg, because we do the config > removal call > > on the node we want to delete. > > > > So gwcli would report success and the ls command will show it > as no > > longer running/exported, but if you restart the rbd-target-api > service > > then it will show up again. > > > > There is an internal command to do what you want. I will post > a PR for > > gwlci and so it can be used by dashboard. > > > > > > On 12/03/2019 01:19 PM, Jason Dillaman wrote: > > > If I recall correctly, the recent ceph-iscsi release > supports the > > > removal of a gateway via the "gwcli". I think the Ceph > dashboard can > > > do that as well. > > > > > > On Tue, Dec 3, 2019 at 1:59 PM Wesley Dillingham > > <wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>> > <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>>>> wrote: > > >> > > >> We utilize 4 iSCSI gateways in a cluster and have noticed the > > following during patching cycles when we sequentially reboot > single > > iSCSI-gateways: > > >> > > >> "gwcli" often hangs on the still-up iSCSI GWs but sometimes > still > > functions and gives the message: > > >> > > >> "1 gateway is inaccessible - updates will be disabled" > > >> > > >> This got me thinking about what the course of action would be > > should an iSCSI gateway fail permanently or semi-permanently, > say a > > hardware issue. What would be the best course of action to > instruct > > the remaining iSCSI gateways that one of them is no longer > available > > and that they should allow updates again and take ownership of the > > now-defunct-node's LUNS? > > >> > > >> I'm guessing pulling down the RADOS config object and rewriting > > it and re-put'ing it followed by a rbd-target-api restart might do > > the trick but am hoping there is a more "in-band" and less > > potentially devastating way to do this. > > >> > > >> Thanks for any insights. > > >> > > >> Respectfully, > > >> > > >> Wes Dillingham > > >> wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>> > <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com> <mailto:wes@wesdillingham.com <mailto:wes@wesdillingham.com>>> > > >> LinkedIn > > >> _______________________________________________ > > >> ceph-users mailing list -- ceph-users@ceph.io <mailto:ceph-users@ceph.io> > <mailto:ceph-users@ceph.io <mailto:ceph-users@ceph.io>> > > <mailto:ceph-users@ceph.io <mailto:ceph-users@ceph.io> <mailto:ceph-users@ceph.io <mailto:ceph-users@ceph.io>>> > > >> To unsubscribe send an email to ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io> > <mailto:ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io>> > > <mailto:ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io> > <mailto:ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io>>> > > > > > > > > > > > > _______________________________________________ > ceph-users mailing list -- ceph-users@ceph.io <mailto:ceph-users@ceph.io> > <mailto:ceph-users@ceph.io <mailto:ceph-users@ceph.io>> > To unsubscribe send an email to ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io> > <mailto:ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io>> > _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io <mailto:ceph-users@ceph.io> To unsubscribe send an email to ceph-users-leave@ceph.io <mailto:ceph-users-leave@ceph.io>
participants (5)
-
Gesiel Galvão Bernardes
-
Jason Dillaman
-
Mike Christie
-
Paul Emmerich
-
Wesley Dillingham