Redeploy iSCSI Gateway fail - 167 returned from docker run
CEPH 16.2.4. I was having an issue where I put a server into maintenance mode and after doing so, the containers for the iSCSI gateway were not running, so I decided to do a redeploy of the service. This caused all the servers running iSCSI to get in a state where it looks like ceph orch was trying to delete the container, but it was stuck. My only recourse was to reboot the servers. I ended up doing a ‘ceph orch rm iscsi.iscsi’ to just remove the services and then tried to redeploy. When I do this, I’m seeing the following in the cephadm logs on the servers where the iscsi gateway is being deployed: 2021-06-01 19:48:15,110 INFO Deploy daemon iscsi.iscsi.cxcto-c240-j27-02.zeypah ... 2021-06-01 19:48:15,111 DEBUG Running command: /bin/docker run --rm --ipc=host --net=host --entrypoint stat --init -e CONTAINER_IMAGE=docker.io/ceph/ceph@sha256:54e95ae1e11404157d7b329d0bef866ebbb214b195a009e87aae4eba9d282949 -e NODE_NAME=cxcto-c240-j27-02.cisco.com -e CEPH_USE_RANDOM_NONCE=1 docker.io/ceph/ceph@sha256:54e95ae1e11404157d7b329d0bef866ebbb214b195a009e87aae4eba9d282949 -c %u %g /var/lib/ceph 2021-06-01 19:48:15,529 DEBUG stat: 167 167 Later in the logs I see: 2021-06-01 19:48:25,933 DEBUG Running command: /bin/docker inspect --format {{.Id}},{{.Config.Image}},{{.Image}},{{.Created}},{{index .Config.Labels "io.ceph.version"}} ceph-a67d529e-ba7f-11eb-940b-5c838f8013a5-iscsi.iscsi.cxcto-c240-j27-02.zeypah 2021-06-01 19:48:25,984 DEBUG /bin/docker: 2021-06-01 19:48:25,984 DEBUG /bin/docker: Error: No such object: ceph-a67d529e-ba7f-11eb-940b-5c838f8013a5-iscsi.iscsi.cxcto-c240-j27-02.zeypah Obviously no such object because the container creation failed. If I try to run that command that is in the logs manually, I get: [root@cxcto-c240-j27-02 ceph]# /bin/docker run --rm --ipc=host --net=host --entrypoint stat --init -e CONTAINER_IMAGE=docker.io/ceph/ceph@sha256:54e95ae1e11404157d7b329d0bef866ebbb214b195a009e87aae4eba9d282949 -e NODE_NAME=cxcto-c240-j27-02.cisco.com -e CEPH_USE_RANDOM_NONCE=1 docker.io/ceph/ceph@sha256:54e95ae1e11404157d7b329d0bef866ebbb214b195a009e87aae4eba9d282949 -c %u %g /var/lib/ceph stat: cannot stat '%g': No such file or directory 167 So the 167 seems to line up with what’s showing up in the script. I’m not clear on what the deal is with the %g. What is supposed to be in that placeholder? Any thoughts on why this is failing? Right now all my iSCSI gateways are down and basically my whole environment is down as a result 🙁 -Paul
Ok - so looks like that 167 was a red herring. That was actually a valid result. The issue is the container was starting up then dying. Looks like this all went downhill as a result of me changing the name of an image (I had named it iscsi-img-005 instead of iscsi-img-0005 to match all my other image names). Looks like iSCSI gateway does not like that. It worked fine after the rename, but when I redeployed the gateways, they never came back up. I saw an error in the logs that indicated it was having trouble with that name. I managed to get things back up and running on one gateway after a lot of messing around getting it to the point where I could delete the configuration that it didn’t like. Right now I’m running on one gateway fine. I tried to scale back up to 4 servers and the other three all have different issues. One comes up, but when I try to provision a target to use it, I get an out of index error. On the 3rd the docker containers come up, but the gateway never shows up. The 4th the containers never come up because it fails with this message: subprocess.CalledProcessError: Command 'ceph -n client.iscsi.iscsi.cxcto-c240-j27-05.noraaw --conf /etc/ceph/ceph.conf osd blacklist rm 10.122.242.200:6977/1317769556' returned non-zero exit status 13. I feel like it would probably be best to just wipe all the iscsi gateway configuration and start from scratch with the iSCSI configuration piece - however even when I remove the service (ceph orch rm iscsi.iscsi) the configuration appears to still be maintained. Where is all this configuration stored? Is there a way to completely remove it to start the iscsi gateways on a clean slate? -Paul
On Jun 1, 2021, at 8:05 PM, Paul Giralt (pgiralt) <pgiralt@cisco.com> wrote:
CEPH 16.2.4. I was having an issue where I put a server into maintenance mode and after doing so, the containers for the iSCSI gateway were not running, so I decided to do a redeploy of the service. This caused all the servers running iSCSI to get in a state where it looks like ceph orch was trying to delete the container, but it was stuck. My only recourse was to reboot the servers. I ended up doing a ‘ceph orch rm iscsi.iscsi’ to just remove the services and then tried to redeploy. When I do this, I’m seeing the following in the cephadm logs on the servers where the iscsi gateway is being deployed:
2021-06-01 19:48:15,110 INFO Deploy daemon iscsi.iscsi.cxcto-c240-j27-02.zeypah ... 2021-06-01 19:48:15,111 DEBUG Running command: /bin/docker run --rm --ipc=host --net=host --entrypoint stat --init -e CONTAINER_IMAGE=docker.io/ceph/ceph@sha256:54e95ae1e11404157d7b329d0bef866ebbb214b195a009e87aae4eba9d282949 -e NODE_NAME=cxcto-c240-j27-02.cisco.com -e CEPH_USE_RANDOM_NONCE=1 docker.io/ceph/ceph@sha256:54e95ae1e11404157d7b329d0bef866ebbb214b195a009e87aae4eba9d282949 -c %u %g /var/lib/ceph 2021-06-01 19:48:15,529 DEBUG stat: 167 167
Later in the logs I see:
2021-06-01 19:48:25,933 DEBUG Running command: /bin/docker inspect --format {{.Id}},{{.Config.Image}},{{.Image}},{{.Created}},{{index .Config.Labels "io.ceph.version"}} ceph-a67d529e-ba7f-11eb-940b-5c838f8013a5-iscsi.iscsi.cxcto-c240-j27-02.zeypah 2021-06-01 19:48:25,984 DEBUG /bin/docker: 2021-06-01 19:48:25,984 DEBUG /bin/docker: Error: No such object: ceph-a67d529e-ba7f-11eb-940b-5c838f8013a5-iscsi.iscsi.cxcto-c240-j27-02.zeypah
Obviously no such object because the container creation failed.
If I try to run that command that is in the logs manually, I get:
[root@cxcto-c240-j27-02 ceph]# /bin/docker run --rm --ipc=host --net=host --entrypoint stat --init -e CONTAINER_IMAGE=docker.io/ceph/ceph@sha256:54e95ae1e11404157d7b329d0bef866ebbb214b195a009e87aae4eba9d282949 -e NODE_NAME=cxcto-c240-j27-02.cisco.com -e CEPH_USE_RANDOM_NONCE=1 docker.io/ceph/ceph@sha256:54e95ae1e11404157d7b329d0bef866ebbb214b195a009e87aae4eba9d282949 -c %u %g /var/lib/ceph stat: cannot stat '%g': No such file or directory 167
So the 167 seems to line up with what’s showing up in the script. I’m not clear on what the deal is with the %g. What is supposed to be in that placeholder? Any thoughts on why this is failing?
Right now all my iSCSI gateways are down and basically my whole environment is down as a result 🙁
-Paul
_______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io
participants (1)
-
Paul Giralt (pgiralt)