Hi Sam,
Yeah somehow `lo:` is not getting skipped, probably due to those
patches. (I guess it is because the 2nd patch looks for `lo:` but in
fact the ifa_name is probably just `lo` without the colon)
https://github.com/ceph/ceph/blob/master/src/common/ipaddr.cc#L110
I don't know why this impacts you but not us -- we already upgraded
one of our clusters to 14.2.18 on Centos 8, and ceph is choosing the
correct interface without needing any network options. And lo: is the
first interface [1] here too.
Could it be as simple as the iface names being sorted alphabetically?
Here we have ens785f0 which would come before lo, but your interface
`p2p2` would come after.
-- dan
[1]
# ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
group default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: eno1: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq state
DOWN group default qlen 1000
link/ether a4:bf:01:60:67:a0 brd ff:ff:ff:ff:ff:ff
3: ens785f0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state
UP group default qlen 1000
link/ether 0c:42:a1:ad:36:9a brd ff:ff:ff:ff:ff:ff
inet 10.116.6.8/26 brd 10.116.6.63 scope global dynamic
noprefixroute ens785f0
valid_lft 432177sec preferred_lft 432177sec
inet6 fd01:1458:e00:1e::100:5/128 scope global dynamic noprefixroute
valid_lft 513502sec preferred_lft 513502sec
inet6 fe80::bdbd:76be:63fd:a4c2/64 scope link noprefixroute
valid_lft forever preferred_lft forever
4: ens785f1: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq
state DOWN group default qlen 1000
link/ether 0c:42:a1:ad:36:9b brd ff:ff:ff:ff:ff:ff
5: eno2: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq state
DOWN group default qlen 1000
link/ether a4:bf:01:60:67:a1 brd ff:ff:ff:ff:ff:ff\
On Mon, Mar 22, 2021 at 8:35 PM Sam Skipsey <aoanla(a)gmail.com> wrote:
>
> Hi Dan:
>
> Aha - I think the first commit is probably it - before that commit, the fact that lo
is highest in the interfaces enumeration didn't matter for us [since it would always
be skipped].
>
> This actually almost certainly also is associated with that other site with a similar
problem (OSDs drop out until you restart the network interface), since I imagine that
would reorder the interface list.
>
> Playing with our public and cluster bind address explicitly does seem to help, so
we'll iterate on that and get to a suitable ceph.conf.
>
> Thanks for the help [and it was the network all along]!
>
>
> Sam
>
> On Mon, 22 Mar 2021 at 19:12, Dan van der Ster <dan(a)vanderster.com> wrote:
>>
>> There are two commits between 14.2.16 and 14.2.18 related to loopback network.
Perhaps one of these is responsible for your issue [1].
>>
>> I'd try playing with the options like cluster/public bind addr and
cluster/public bind interface until you can convince the osd to bind to the correct
listening IP.
>>
>> (That said, i don't know which version you're running on the logs shared
earlier. But I think you should try to get 14.2.18 working anyway).
>>
>> .. dan
>>
>> [1]
>>
>> > git log v14.2.18...v14.2.16 ipaddr.cc commit
89321762ad4cfdd1a68cae467181bdd1a501f14d
>> Author: Thomas Goirand <zigo(a)debian.org>
>> Date: Fri Jan 15 10:50:05 2021 +0100
>>
>> common/ipaddr: Allow binding on lo
>>
>> Commmit 5cf0fa872231f4eaf8ce6565a04ed675ba5b689b, solves the issue that
>> the osd can't restart after seting a virtual local loopback IP. However,
>> this commit also prevents a bgp-to-the-host over unumbered Ipv6
>> local-link is setup, where OSD typically are bound to the lo interface.
>>
>> To solve this, this single char patch simply checks against "lo:"
to
>> match only virtual interfaces instead of anything that starts with
"lo".
>>
>> Fixes:
https://tracker.ceph.com/issues/48893
>> Signed-off-by: Thomas Goirand <zigo(a)debian.org>
>> (cherry picked from commit 201b59204374ebdab91bb554b986577a97b19c36)
>>
>> commit b52cae90d67eb878b3ddfe547b8bf16e0d4d1a45
>> Author: lijaiwei1 <lijiawei1(a)chinatelecom.cn>
>> Date: Tue Dec 24 22:34:46 2019 +0800
>>
>> common: skip interfaces starting with "lo" in
find_ipv{4,6}_in_subnet()
>>
>> This will solve the issue that the osd can't restart after seting a
>> virtual local loopback IP.
>> In find_ipv4_in_subnet() and find_ipv6_in_subnet(), I use
>> boost::starts_with(addrs->ifa_name, "lo") to ship the
interfaces
>> starting with "lo".
>>
>> Fixes:
https://tracker.ceph.com/issues/43417
>> Signed-off-by: Jiawei Li <lijiawei1(a)chinatelecom.cn>
>> (cherry picked from commit 5cf0fa872231f4eaf8ce6565a04ed675ba5b689b)
>>
>>
>>
>>
>>
>> On Mon, Mar 22, 2021, 7:42 PM Sam Skipsey <aoanla(a)gmail.com> wrote:
>>>
>>> I don't think we explicitly set any ms settings in the OSD host ceph.conf
[all the OSDs ceph.confs are identical across the entire cluster].
>>>
>>> ip a gives:
>>>
>>> ip a
>>> 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
group default qlen 1000
>>> link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
>>> inet 127.0.0.1/8 scope host lo
>>> valid_lft forever preferred_lft forever
>>> inet6 ::1/128 scope host
>>> valid_lft forever preferred_lft forever
>>> 2: em1: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq state
DOWN group default qlen 1000
>>> link/ether 4c:d9:8f:55:92:f6 brd ff:ff:ff:ff:ff:ff
>>> 3: em2: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq state
DOWN group default qlen 1000
>>> link/ether 4c:d9:8f:55:92:f7 brd ff:ff:ff:ff:ff:ff
>>> 4: p2p1: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq state
DOWN group default qlen 1000
>>> link/ether b4:96:91:3f:62:20 brd ff:ff:ff:ff:ff:ff
>>> 5: p2p2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP
group default qlen 1000
>>> link/ether b4:96:91:3f:62:22 brd ff:ff:ff:ff:ff:ff
>>> inet 10.1.50.21/8 brd 10.255.255.255 scope global noprefixroute p2p2
>>> valid_lft forever preferred_lft forever
>>> inet6 fe80::b696:91ff:fe3f:6222/64 scope link noprefixroute
>>> valid_lft forever preferred_lft forever
>>>
>>> (where here p2p2 is the only active network link, and is also the private and
public network for the ceph cluster)
>>>
>>> The output is similar on other hosts - with p2p2 either at position 3 or 5
depending on the order the interfaces were enumerated.
>>>
>>> Sam
>>>
>>> On Mon, 22 Mar 2021 at 17:34, Dan van der Ster <dan(a)vanderster.com>
wrote:
>>>>
>>>> Which `ms` settings do you have in the OSD host's ceph.conf or the
ceph config dump?
>>>>
>>>> And how does `ip a` look on one of these hosts where the osd is
registering itself as 127.0.0.1?
>>>>
>>>>
>>>> You might as well set nodown again now. This will make ops pile up, but
that's the least of your concerns at the moment.
>>>> (With osds flapping the osdmaps churn and that inflates the mon store)
>>>>
>>>> .. Dan
>>>>
>>>> On Mon, Mar 22, 2021, 6:28 PM Sam Skipsey <aoanla(a)gmail.com>
wrote:
>>>>>
>>>>> Hm, yes it does [and I was wondering why loopbacks were showing up
suddenly in the logs]. This wasn't happening with 14.2.16 so what's changed about
how we specify stuff?
>>>>>
>>>>> This might correlate with the other person on the IRC list who has
problems with 14.2.18 and their OSDs deciding they don't work sometimes until they
forcibly restart their network links...
>>>>>
>>>>>
>>>>> Sam
>>>>>
>>>>> On Mon, 22 Mar 2021 at 17:20, Dan van der Ster
<dan(a)vanderster.com> wrote:
>>>>>>
>>>>>> What's with the OSDs having loopback addresses? E.g.
v2:127.0.0.1:6881/17664667,v1:127.0.0.1:6882/17664667
>>>>>>
>>>>>> Does `ceph osd dump` show those same loopback addresses for each
OSD?
>>>>>>
>>>>>> This sounds familiar... I'm trying to find the recent
ticket.
>>>>>>
>>>>>> .. dan
>>>>>>
>>>>>>
>>>>>> On Mon, Mar 22, 2021, 6:07 PM Sam Skipsey
<aoanla(a)gmail.com> wrote:
>>>>>>>
>>>>>>> hi Dan:
>>>>>>>
>>>>>>> So, unsetting nodown results in... almost all of the OSDs
being marked down. (231 down out of 328).
>>>>>>> Checking the actual OSD services, most of them were actually
up and active on the nodes, even when the mons had marked them down.
>>>>>>> (On a few nodes, the down services corresponded to OSDs that
had been flapping - but increasing osd_max_markdown locally to keep them up despite the
previous flapping, and restarting the services... didn't help.)
>>>>>>>
>>>>>>> In fact, starting up the few OSD services which had actually
stopped, resulted in a different set of OSDs being marked down, and some others coming
up.
>>>>>>> We currently have a sort of "rolling OSD outness"
passing through the cluster - there's always ~230 OSDs marked down now, but which ones
those are changes (we've had everything from 1 HOST down to 4 HOSTS down over the past
14 minutes as things fluctuate.
>>>>>>>
>>>>>>> A log from one of the "down" OSDs [which is
actually running, and on the same host as OSDs which are marked up] shows this worrying
snippet
>>>>>>>
>>>>>>> 2021-03-22 17:01:45.298 7f6c9c883700 1 osd.127 253515
is_healthy false -- only 0/10 up peers (less than 33%)
>>>>>>> 2021-03-22 17:01:45.298 7f6c9c883700 1 osd.127 253515 not
healthy; waiting to boot
>>>>>>> 2021-03-22 17:01:46.340 7f6c9c883700 1 osd.127 253515
is_healthy false -- only 0/10 up peers (less than 33%)
>>>>>>> 2021-03-22 17:01:46.340 7f6c9c883700 1 osd.127 253515 not
healthy; waiting to boot
>>>>>>> 2021-03-22 17:01:47.376 7f6c9c883700 1 osd.127 253515
is_healthy false -- only 0/10 up peers (less than 33%)
>>>>>>> 2021-03-22 17:01:47.376 7f6c9c883700 1 osd.127 253515 not
healthy; waiting to boot
>>>>>>> 2021-03-22 17:01:48.395 7f6c9c883700 1 osd.127 253515
is_healthy false -- only 0/10 up peers (less than 33%)
>>>>>>> 2021-03-22 17:01:48.395 7f6c9c883700 1 osd.127 253515 not
healthy; waiting to boot
>>>>>>> 2021-03-22 17:01:49.407 7f6c9c883700 1 osd.127 253515
is_healthy false -- only 0/10 up peers (less than 33%)
>>>>>>> 2021-03-22 17:01:49.407 7f6c9c883700 1 osd.127 253515 not
healthy; waiting to boot
>>>>>>> 2021-03-22 17:01:50.400 7f6c9c883700 1 osd.127 253515
is_healthy false -- only 0/10 up peers (less than 33%)
>>>>>>> 2021-03-22 17:01:50.400 7f6c9c883700 1 osd.127 253515 not
healthy; waiting to boot
>>>>>>> 2021-03-22 17:01:50.922 7f6c9f088700 -1 --2-
10.1.50.21:0/23673 >> [v2:127.0.0.1:6881/17664667,v1:127.0.0.1:6882/17664667]
conn(0x56010903e400 0x56011a71fc00 unknown :-1 s=BANNER_CONNECTING pgs=0 cs=0 l=1 rev1=0
rx=0 tx=0)._handle_peer_banner peer
[v2:127.0.0.1:6881/17664667,v1:127.0.0.1:6882/17664667] is using msgr V1 protocol
>>>>>>> 2021-03-22 17:01:50.922 7f6c9f889700 -1 --2-
10.1.50.21:0/23673 >> [v2:127.0.0.1:6821/13015214,v1:127.0.0.1:6831/13015214]
conn(0x5600df434000 0x56011718e000 unknown :-1 s=BANNER_CONNECTING pgs=0 cs=0 l=1 rev1=0
rx=0 tx=0)._handle_peer_banner peer
[v2:127.0.0.1:6821/13015214,v1:127.0.0.1:6831/13015214] is using msgr V1 protocol
>>>>>>> 2021-03-22 17:01:50.922 7f6ca008a700 -1 --2-
10.1.50.21:0/23673 >> [v2:127.0.0.1:6826/11091658,v1:127.0.0.1:6828/11091658]
conn(0x5600f85ed800 0x560109df2a00 unknown :-1 s=BANNER_CONNECTING pgs=0 cs=0 l=1 rev1=0
rx=0 tx=0)._handle_peer_banner peer
[v2:127.0.0.1:6826/11091658,v1:127.0.0.1:6828/11091658] is using msgr V1 protocol
>>>>>>> 2021-03-22 17:01:50.922 7f6ca008a700 -1 --2-
10.1.50.21:0/23673 >> [v2:127.0.0.1:6859/2683393,v1:127.0.0.1:6862/2683393]
conn(0x5600f22ea000 0x560117182300 unknown :-1 s=BANNER_CONNECTING pgs=0 cs=0 l=1 rev1=0
rx=0 tx=0)._handle_peer_banner peer [v2:127.0.0.1:6859/2683393,v1:127.0.0.1:6862/2683393]
is using msgr V1 protocol
>>>>>>> 2021-03-22 17:01:50.922 7f6ca008a700 -1 --2-
10.1.50.21:0/23673 >> [v2:127.0.0.1:6901/15090566,v1:127.0.0.1:6907/15090566]
conn(0x5600df435c00 0x560139370300 unknown :-1 s=BANNER_CONNECTING pgs=0 cs=0 l=1 rev1=0
rx=0 tx=0)._handle_peer_banner peer
[v2:127.0.0.1:6901/15090566,v1:127.0.0.1:6907/15090566] is using msgr V1 protocol
>>>>>>> 2021-03-22 17:01:51.377 7f6c9c883700 1 osd.127 253515
is_healthy false -- only 0/10 up peers (less than 33%)
>>>>>>> 2021-03-22 17:01:51.377 7f6c9c883700 1 osd.127 253515 not
healthy; waiting to boot
>>>>>>> 2021-03-22 17:01:52.370 7f6c9c883700 1 osd.127 253515
is_healthy false -- only 0/10 up peers (less than 33%)
>>>>>>> 2021-03-22 17:01:52.370 7f6c9c883700 1 osd.127 253515 not
healthy; waiting to boot
>>>>>>> 2021-03-22 17:01:53.377 7f6c9c883700 1 osd.127 253515
is_healthy false -- only 0/10 up peers (less than 33%)
>>>>>>> 2021-03-22 17:01:53.377 7f6c9c883700 1 osd.127 253515 not
healthy; waiting to boot
>>>>>>> 2021-03-22 17:01:54.385 7f6c9c883700 1 osd.127 253515
is_healthy false -- only 0/10 up peers (less than 33%)
>>>>>>> 2021-03-22 17:01:54.385 7f6c9c883700 1 osd.127 253515 not
healthy; waiting to boot
>>>>>>> 2021-03-22 17:01:55.385 7f6c9c883700 1 osd.127 253515
is_healthy false -- only 0/10 up peers (less than 33%)
>>>>>>> 2021-03-22 17:01:55.385 7f6c9c883700 1 osd.127 253515 not
healthy; waiting to boot
>>>>>>> 2021-03-22 17:01:56.362 7f6c9c883700 1 osd.127 253515
is_healthy false -- only 0/10 up peers (less than 33%)
>>>>>>> 2021-03-22 17:01:56.362 7f6c9c883700 1 osd.127 253515 not
healthy; waiting to boot
>>>>>>> 2021-03-22 17:01:57.324 7f6c9c883700 1 osd.127 253515
is_healthy false -- only 0/10 up peers (less than 33%)
>>>>>>> 2021-03-22 17:01:57.324 7f6c9c883700 1 osd.127 253515 not
healthy; waiting to boot
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Any suggestions?
>>>>>>>
>>>>>>> Sam
>>>>>>>
>>>>>>> P.S. an example ceph status as it is now [with everything now
on 14.2.18, since we had to restart osds anyway]:
>>>>>>>
>>>>>>> cluster:
>>>>>>> id: a1148af2-6eaf-4486-a27e-a05a78c2b378
>>>>>>> health: HEALTH_WARN
>>>>>>> pauserd,pausewr,noout,nobackfill,norebalance
flag(s) set
>>>>>>> 230 osds down
>>>>>>> 4 hosts (80 osds) down
>>>>>>> Reduced data availability: 2048 pgs inactive
>>>>>>> 8 slow ops, oldest one blocked for 901 sec,
mon.cephs01 has slow ops
>>>>>>>
>>>>>>> services:
>>>>>>> mon: 3 daemons, quorum cephs01,cephs02,cephs03 (age 2h)
>>>>>>> mgr: cephs01(active, since 77m)
>>>>>>> osd: 329 osds: 98 up (since 4s), 328 in (since 4d)
>>>>>>> flags pauserd,pausewr,noout,nobackfill,norebalance
>>>>>>>
>>>>>>> data:
>>>>>>> pools: 3 pools, 2048 pgs
>>>>>>> objects: 0 objects, 0 B
>>>>>>> usage: 0 B used, 0 B / 0 B avail
>>>>>>> pgs: 100.000% pgs unknown
>>>>>>> 2048 unknown
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On Mon, 22 Mar 2021 at 14:57, Dan van der Ster
<dan(a)vanderster.com> wrote:
>>>>>>>>
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> I would unset nodown (hiding osd failures) and norecover
(blcoking PGs
>>>>>>>> from recovering degraded objects), then start starting
osds.
>>>>>>>> As soon as you have some osd logs reporting some
failures, then share those...
>>>>>>>>
>>>>>>>> - Dan
>>>>>>>>
>>>>>>>> On Mon, Mar 22, 2021 at 3:49 PM Sam Skipsey
<aoanla(a)gmail.com> wrote:
>>>>>>>> >
>>>>>>>> > So, we started the mons and mgr up again, and
here's the relevant logs, including also ceph versions. We've also turned off all
of the firewalls on all of the nodes so we know that there can't be network issues
[and, indeed, all of our management of the OSDs happens via logins from the service nodes
or to each other]
>>>>>>>> >
>>>>>>>> > > ceph status
>>>>>>>> >
>>>>>>>> >
>>>>>>>> > cluster:
>>>>>>>> > id: a1148af2-6eaf-4486-a27e-a05a78c2b378
>>>>>>>> > health: HEALTH_WARN
>>>>>>>> >
pauserd,pausewr,nodown,noout,nobackfill,norebalance,norecover flag(s) set
>>>>>>>> > 1 nearfull osd(s)
>>>>>>>> > 3 pool(s) nearfull
>>>>>>>> > Reduced data availability: 2048 pgs
inactive
>>>>>>>> > mons cephs01,cephs02,cephs03 are using a
lot of disk space
>>>>>>>> >
>>>>>>>> > services:
>>>>>>>> > mon: 3 daemons, quorum cephs01,cephs02,cephs03
(age 61s)
>>>>>>>> > mgr: cephs01(active, since 76s)
>>>>>>>> > osd: 329 osds: 329 up (since 63s), 328 in (since
4d); 466 remapped pgs
>>>>>>>> > flags
pauserd,pausewr,nodown,noout,nobackfill,norebalance,norecover
>>>>>>>> >
>>>>>>>> > data:
>>>>>>>> > pools: 3 pools, 2048 pgs
>>>>>>>> > objects: 0 objects, 0 B
>>>>>>>> > usage: 0 B used, 0 B / 0 B avail
>>>>>>>> > pgs: 100.000% pgs unknown
>>>>>>>> > 2048 unknown
>>>>>>>> >
>>>>>>>> >
>>>>>>>> > > ceph health detail
>>>>>>>> >
>>>>>>>> > HEALTH_WARN
pauserd,pausewr,nodown,noout,nobackfill,norebalance,norecover flag(s) set; 1 nearfull
osd(s); 3 pool(s) nearfull; Reduced data availability: 2048 pgs inactive; mons
cephs01,cephs02,cephs03 are using a lot of disk space
>>>>>>>> > OSDMAP_FLAGS
pauserd,pausewr,nodown,noout,nobackfill,norebalance,norecover flag(s) set
>>>>>>>> > OSD_NEARFULL 1 nearfull osd(s)
>>>>>>>> > osd.63 is near full
>>>>>>>> > POOL_NEARFULL 3 pool(s) nearfull
>>>>>>>> > pool 'dteam' is nearfull
>>>>>>>> > pool 'atlas' is nearfull
>>>>>>>> > pool 'atlas-localgroup' is nearfull
>>>>>>>> > PG_AVAILABILITY Reduced data availability: 2048 pgs
inactive
>>>>>>>> > pg 13.1ef is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 13.1f0 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 13.1f1 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 13.1f2 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 13.1f3 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 13.1f4 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 13.1f5 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 13.1f6 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 13.1f7 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 13.1f8 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 13.1f9 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 13.1fa is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 13.1fb is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 13.1fc is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 13.1fd is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 13.1fe is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 13.1ff is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1ec is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1f0 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1f1 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1f2 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1f3 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1f4 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1f5 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1f6 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1f7 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1f8 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1f9 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1fa is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1fb is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1fc is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1fd is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1fe is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 14.1ff is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1ed is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1f0 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1f1 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1f2 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1f3 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1f4 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1f5 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1f6 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1f7 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1f8 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1f9 is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1fa is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1fb is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1fc is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1fd is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1fe is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > pg 15.1ff is stuck inactive for 89.322981,
current state unknown, last acting []
>>>>>>>> > MON_DISK_BIG mons cephs01,cephs02,cephs03 are using
a lot of disk space
>>>>>>>> > mon.cephs01 is 96 GiB >= mon_data_size_warn
(15 GiB)
>>>>>>>> > mon.cephs02 is 96 GiB >= mon_data_size_warn
(15 GiB)
>>>>>>>> > mon.cephs03 is 96 GiB >= mon_data_size_warn
(15 GiB)
>>>>>>>> >
>>>>>>>> >
>>>>>>>> > > ceph versions
>>>>>>>> >
>>>>>>>> > {
>>>>>>>> > "mon": {
>>>>>>>> > "ceph version 14.2.18
(befbc92f3c11eedd8626487211d200c0b44786d9) nautilus (stable)": 3
>>>>>>>> > },
>>>>>>>> > "mgr": {
>>>>>>>> > "ceph version 14.2.18
(befbc92f3c11eedd8626487211d200c0b44786d9) nautilus (stable)": 1
>>>>>>>> > },
>>>>>>>> > "osd": {
>>>>>>>> > "ceph version 14.2.10
(b340acf629a010a74d90da5782a2c5fe0b54ac20) nautilus (stable)": 1,
>>>>>>>> > "ceph version 14.2.15
(afdd217ae5fb1ed3f60e16bd62357ca58cc650e5) nautilus (stable)": 188,
>>>>>>>> > "ceph version 14.2.16
(762032d6f509d5e7ee7dc008d80fe9c87086603c) nautilus (stable)": 18,
>>>>>>>> > "ceph version 14.2.18
(befbc92f3c11eedd8626487211d200c0b44786d9) nautilus (stable)": 122
>>>>>>>> > },
>>>>>>>> >
>>>>>>>> >
>>>>>>>> > >>>>>>
>>>>>>>> >
>>>>>>>> > As a note, the log where the mgr explodes (which
precipitated all of this) definitely shows the problem occurring on the 12th [when 14.2.17
dropped], but things didn't "break" until we tried upgrading OSDs to
14.2.18...
>>>>>>>> >
>>>>>>>> >
>>>>>>>> > Sam
>>>>>>>> >
>>>>>>>> >
>>>>>>>> > On Mon, 22 Mar 2021 at 12:20, Sam Skipsey
<aoanla(a)gmail.com> wrote:
>>>>>>>> >>
>>>>>>>> >> Hi Dan:
>>>>>>>> >>
>>>>>>>> >> Thanks for the reply - at present, our mons and
mgrs are off [because of the unsustainable nature of the filesystem usage]. We'll try
putting them on again for long enough to get "ceph status" out of them, but
because the mgr was unable to actually talk to anything, and reply at that point.
>>>>>>>> >>
>>>>>>>> >> (And thanks for the link to the bug tracker - I
guess this mismatch of expectations is why the devs are so keen to move to containerised
deployments where there is no co-location of different types of server, as it means they
don't need to worry as much about the assumptions about when it's okay to restart
a service on package update. Disappointing that it seems stale after 2 years...)
>>>>>>>> >>
>>>>>>>> >> Sam
>>>>>>>> >>
>>>>>>>> >>
>>>>>>>> >>
>>>>>>>> >> On Mon, 22 Mar 2021 at 12:11, Dan van der Ster
<dan(a)vanderster.com> wrote:
>>>>>>>> >>>
>>>>>>>> >>> Hi Sam,
>>>>>>>> >>>
>>>>>>>> >>> The daemons restart (for *some* releases)
because of this:
>>>>>>>> >>>
https://tracker.ceph.com/issues/21672
>>>>>>>> >>> In short, if the selinux module changes, and
if you have selinux
>>>>>>>> >>> enabled, then midway through yum update,
there will be a systemctl
>>>>>>>> >>> restart ceph.target issued.
>>>>>>>> >>>
>>>>>>>> >>> For the rest -- I think you should focus on
getting the PGs all
>>>>>>>> >>> active+clean as soon as possible, because
the degraded and remapped
>>>>>>>> >>> states are what leads to mon / osdmap
growth.
>>>>>>>> >>> This kind of scenario is why we wrote this
tool:
>>>>>>>> >>>
https://github.com/cernceph/ceph-scripts/blob/master/tools/upmap/upmap-rema…
>>>>>>>> >>> It will use pg-upmap-items to force the PGs
to the OSDs where they are
>>>>>>>> >>> currently residing.
>>>>>>>> >>>
>>>>>>>> >>> But there is some clarification needed
before you go ahead with that.
>>>>>>>> >>> Could you share `ceph status`, `ceph health
detail`?
>>>>>>>> >>>
>>>>>>>> >>> Cheers, Dan
>>>>>>>> >>>
>>>>>>>> >>>
>>>>>>>> >>> On Mon, Mar 22, 2021 at 12:05 PM Sam Skipsey
<aoanla(a)gmail.com> wrote:
>>>>>>>> >>> >
>>>>>>>> >>> > Hi everyone:
>>>>>>>> >>> >
>>>>>>>> >>> > I posted to the list on Friday morning
(UK time), but apparently my email
>>>>>>>> >>> > is still in moderation (I have an email
from the list bot telling me that
>>>>>>>> >>> > it's held for moderation but no
updates).
>>>>>>>> >>> >
>>>>>>>> >>> > Since this is a bit urgent - we have
~3PB of storage offline - I'm posting
>>>>>>>> >>> > again.
>>>>>>>> >>> >
>>>>>>>> >>> > To save retyping the whole thing, I
will direct you to a copy of the email
>>>>>>>> >>> > I wrote on Friday:
>>>>>>>> >>> >
>>>>>>>> >>> >
http://aoanla.pythonanywhere.com/Logs/EmailToCephUsers.txt
>>>>>>>> >>> >
>>>>>>>> >>> > (Since that was sent, we did
successfully add big SSDs to the MON hosts so
>>>>>>>> >>> > they don't fill up their disks with
store.db s).
>>>>>>>> >>> >
>>>>>>>> >>> > I would appreciate any advice -
assuming this also doesn't get stuck in
>>>>>>>> >>> > moderation queues.
>>>>>>>> >>> >
>>>>>>>> >>> > --
>>>>>>>> >>> > Sam Skipsey (he/him, they/them)
>>>>>>>> >>> >
_______________________________________________
>>>>>>>> >>> > ceph-users mailing list --
ceph-users(a)ceph.io
>>>>>>>> >>> > To unsubscribe send an email to
ceph-users-leave(a)ceph.io
>>>>>>>> >>
>>>>>>>> >>
>>>>>>>> >>
>>>>>>>> >> --
>>>>>>>> >> Sam Skipsey (he/him, they/them)
>>>>>>>> >>
>>>>>>>> >>
>>>>>>>> >
>>>>>>>> >
>>>>>>>> > --
>>>>>>>> > Sam Skipsey (he/him, they/them)
>>>>>>>> >
>>>>>>>> >
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> --
>>>>>>> Sam Skipsey (he/him, they/them)
>>>>>>>
>>>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> Sam Skipsey (he/him, they/them)
>>>>>
>>>>>
>>>
>>>
>>> --
>>> Sam Skipsey (he/him, they/them)
>>>
>>>
>
>
> --
> Sam Skipsey (he/him, they/them)
>
>