Hello Eugen, yes, we followed the documentation and everything worked fine. The cache is gone. Removing the pool worked well. Everything is clean. The PGs are empty active+clean. Possible solutions: 1. ceph pg {pg-id} mark_unfound_lost delete I do not think this is the right way since it is for PGs with status unfound. But it could work also. 2. Set the following for the three disk: ceph osd lost {osd-id} I am not sure how the cluster will react to this. 3. ceph-objectstore-tool --data-path /path/to/osd --op remove --pgid 3.0 --force Now, will the cluster accept the removed PG status? 4. The three disks are still presented in the crush rule, class ssd, each single OSD under one host entry. What if I remove them from crush? Do you have a better idea, Eugen? Best, Malte Am 04.10.23 um 09:21 schrieb Eugen Block: > Hi, > > just for clarity, you're actually talking about the cache tier as > described in the docs [1]? And you followed the steps until 'ceph osd > tier remove cold-storage hot-storage' successfully? And the pool has > been really deleted successfully ('ceph osd pool ls detail')? > > [1] > https://docs.ceph.com/en/latest/rados/operations/cache-tiering/#removing-a-… > > Zitat von Malte Stroem <malte.stroem(a)gmail.com>om>: > >> Hello, >> >> we removed an SSD cache tier and its pool. >> >> The PGs for the pool do still exist. >> >> The cluster is healthy. >> >> The PGs are empty and they reside on the cache tier pool's SSDs. >> >> We like to take out the disks but it is not possible. The cluster sees >> the PGs and answers with a HEALTH_WARN. >> >> Because of the replication of three there are still 128 PGs on three >> of the 24 OSDs. We were able to remove the other OSDs. >> >> Summary: >> >> - pool removed >> - 3 x 128 empty PGs still exist >> - 3 of 24 OSDs still exist >> >> How is it possible to remove these empty and healthy PGs? >> >> The only way I found was something like: >> >> ceph pg {pg-id} mark_unfound_lost delete >> >> Is that the right way? >> >> Some output of: >> >> ceph pg ls-by-osd 23 >> >> PG      OBJECTS  DEGRADED  MISPLACED  UNFOUND  BYTES  OMAP_BYTES* >> OMAP_KEYS*  LOG   STATE         SINCE  VERSION         REPORTED UP >>               ACTING             SCRUB_STAMP DEEP_SCRUB_STAMP >> 3.0          0         0          0        0      0            0   0 >>     0  active+clean    27h             0'0    2627265:196316 >> [15,6,23]p15  [15,6,23]p15  2023-09-28T12:41:52.982955+0200 >> 2023-09-27T06:48:23.265838+0200 >> 3.1          0         0          0        0      0            0   0 >>     0  active+clean     9h             0'0    2627266:19330 >> [6,23,15]p6  [6,23,15]p6  2023-09-29T06:30:57.630016+0200 >> 2023-09-27T22:58:21.992451+0200 >> 3.2          0         0          0        0      0            0   0 >>     0  active+clean     2h             0'0   2627265:1135185 >> [23,15,6]p23  [23,15,6]p23  2023-09-29T13:42:07.346658+0200 >> 2023-09-24T14:31:52.844427+0200 >> 3.3          0         0          0        0      0            0   0 >>     0  active+clean    13h             0'0    2627266:193170 >> [6,15,23]p6  [6,15,23]p6  2023-09-29T01:56:54.517337+0200 >> 2023-09-27T17:47:24.961279+0200 >> 3.4          0         0          0        0      0            0   0 >>     0  active+clean    14h             0'0   2627265:2343551 >> [23,6,15]p23  [23,6,15]p23  2023-09-29T00:47:47.548860+0200 >> 2023-09-25T09:39:51.259304+0200 >> 3.5          0         0          0        0      0            0   0 >>     0  active+clean     2h             0'0    2627265:194111 >> [15,6,23]p15  [15,6,23]p15  2023-09-29T13:28:48.879959+0200 >> 2023-09-26T15:35:44.217302+0200 >> 3.6          0         0          0        0      0            0   0 >>     0  active+clean     6h             0'0   2627265:2345717 >> [23,15,6]p23  [23,15,6]p23  2023-09-29T09:26:02.534825+0200 >> 2023-09-27T21:56:57.500126+0200 >> >> Best regards, >> Malte >> _______________________________________________ >> ceph-users mailing list -- ceph-users(a)ceph.io >> To unsubscribe send an email to ceph-users-leave(a)ceph.io > > > _______________________________________________ > ceph-users mailing list -- ceph-users(a)ceph.io > To unsubscribe send an email to ceph-users-leave(a)ceph.io

@Eugen We have seen the same problems 8 years ago. I can only recommend never to use cache tiering in production. At Cephalocon this was part of my talk and as far as I remember cache tiering will also disappear from ceph soon. Cache tiering has been deprecated in the Reef release as it has lacked a maintainer for a very long time. This does not mean it will be certainly removed, but we may choose to remove it without much further notice. https://docs.ceph.com/en/latest/rados/operations/cache-tiering/ Regards, Joachim Am 05.10.23 um 10:02 schrieb Eugen Block: > Which ceph version is this? I'm trying to understand how removing a > pool leaves the PGs of that pool... Do you have any logs or something > from when you removed the pool? > We'll have to deal with a cache tier in the forseeable future as well > so this is quite relevant for us as well. Maybe I'll try to reproduce > it in a test cluster first. > Are those SSDs exclusively for the cache tier or are they used by > other pools as well? If they were used only for the cache tier you > should be able to just remove them without any risk. But as I said, > I'd rather try to understand before purging them. > > > Zitat von Malte Stroem <malte.stroem(a)gmail.com>om>: > >> Hello Eugen, >> >> yes, we followed the documentation and everything worked fine. The >> cache is gone. >> >> Removing the pool worked well. Everything is clean. >> >> The PGs are empty active+clean. >> >> Possible solutions: >> >> 1. >> >> ceph pg {pg-id} mark_unfound_lost delete >> >> I do not think this is the right way since it is for PGs with status >> unfound. But it could work also. >> >> 2. >> >> Set the following for the three disk: >> >> ceph osd lost {osd-id} >> >> I am not sure how the cluster will react to this. >> >> 3. >> >> ceph-objectstore-tool --data-path /path/to/osd --op remove --pgid 3.0 >> --force >> >> Now, will the cluster accept the removed PG status? >> >> 4. >> >> The three disks are still presented in the crush rule, class ssd, >> each single OSD under one host entry. >> >> What if I remove them from crush? >> >> Do you have a better idea, Eugen? >> >> Best, >> Malte >> >> Am 04.10.23 um 09:21 schrieb Eugen Block: >>> Hi, >>> >>> just for clarity, you're actually talking about the cache tier as >>> described in the docs [1]? And you followed the steps until 'ceph >>> osd tier remove cold-storage hot-storage' successfully? And the pool >>> has been really deleted successfully ('ceph osd pool ls detail')? >>> >>> [1] >>> https://docs.ceph.com/en/latest/rados/operations/cache-tiering/#removing-a-… >>> >>> Zitat von Malte Stroem <malte.stroem(a)gmail.com>om>: >>> >>>> Hello, >>>> >>>> we removed an SSD cache tier and its pool. >>>> >>>> The PGs for the pool do still exist. >>>> >>>> The cluster is healthy. >>>> >>>> The PGs are empty and they reside on the cache tier pool's SSDs. >>>> >>>> We like to take out the disks but it is not possible. The cluster >>>> sees the PGs and answers with a HEALTH_WARN. >>>> >>>> Because of the replication of three there are still 128 PGs on >>>> three of the 24 OSDs. We were able to remove the other OSDs. >>>> >>>> Summary: >>>> >>>> - pool removed >>>> - 3 x 128 empty PGs still exist >>>> - 3 of 24 OSDs still exist >>>> >>>> How is it possible to remove these empty and healthy PGs? >>>> >>>> The only way I found was something like: >>>> >>>> ceph pg {pg-id} mark_unfound_lost delete >>>> >>>> Is that the right way? >>>> >>>> Some output of: >>>> >>>> ceph pg ls-by-osd 23 >>>> >>>> PG      OBJECTS  DEGRADED  MISPLACED  UNFOUND  BYTES OMAP_BYTES* >>>> OMAP_KEYS*  LOG   STATE         SINCE VERSION         REPORTED UP >>>> ACTING             SCRUB_STAMP DEEP_SCRUB_STAMP >>>> 3.0          0         0          0        0 0            0   0     >>>> 0  active+clean    27h 0'0    2627265:196316 [15,6,23]p15  >>>> [15,6,23]p15 2023-09-28T12:41:52.982955+0200 >>>> 2023-09-27T06:48:23.265838+0200 >>>> 3.1          0         0          0        0 0            0   0     >>>> 0  active+clean     9h 0'0    2627266:19330 [6,23,15]p6  >>>> [6,23,15]p6 2023-09-29T06:30:57.630016+0200 >>>> 2023-09-27T22:58:21.992451+0200 >>>> 3.2          0         0          0        0 0            0   0     >>>> 0  active+clean     2h 0'0   2627265:1135185 [23,15,6]p23  >>>> [23,15,6]p23 2023-09-29T13:42:07.346658+0200 >>>> 2023-09-24T14:31:52.844427+0200 >>>> 3.3          0         0          0        0 0            0   0     >>>> 0  active+clean    13h 0'0    2627266:193170 [6,15,23]p6  >>>> [6,15,23]p6 2023-09-29T01:56:54.517337+0200 >>>> 2023-09-27T17:47:24.961279+0200 >>>> 3.4          0         0          0        0 0            0   0     >>>> 0  active+clean    14h 0'0   2627265:2343551 [23,6,15]p23  >>>> [23,6,15]p23 2023-09-29T00:47:47.548860+0200 >>>> 2023-09-25T09:39:51.259304+0200 >>>> 3.5          0         0          0        0 0            0   0     >>>> 0  active+clean     2h 0'0    2627265:194111 [15,6,23]p15  >>>> [15,6,23]p15 2023-09-29T13:28:48.879959+0200 >>>> 2023-09-26T15:35:44.217302+0200 >>>> 3.6          0         0          0        0 0            0   0     >>>> 0  active+clean     6h 0'0   2627265:2345717 [23,15,6]p23  >>>> [23,15,6]p23 2023-09-29T09:26:02.534825+0200 >>>> 2023-09-27T21:56:57.500126+0200 >>>> >>>> Best regards, >>>> Malte >>>> _______________________________________________ >>>> ceph-users mailing list -- ceph-users(a)ceph.io >>>> To unsubscribe send an email to ceph-users-leave(a)ceph.io >>> >>> >>> _______________________________________________ >>> ceph-users mailing list -- ceph-users(a)ceph.io >>> To unsubscribe send an email to ceph-users-leave(a)ceph.io > > > _______________________________________________ > ceph-users mailing list -- ceph-users(a)ceph.io > To unsubscribe send an email to ceph-users-leave(a)ceph.io

I know, I know... but since we are already using it (for years) I have to check how to remove it safely, maybe as long as we're on Pacific. ;-) Zitat von Joachim Kraftmayer - ceph ambassador <joachim.kraftmayer(a)clyso.com>om>:

Maybe the mismatching OSD versions had an impact on the unclean tier removal, but this is just a guess. I couldn't reproduce it in a Pacific test cluster, the removal worked fine without leaving behind empty PGs. But I had only a few rbd images in that pool so it's not really representative, I guess. Hopefully Joachim has some more details how they got rid of this situation. Zitat von Malte Stroem <malte.stroem(a)gmail.com>om>:

Hello, well yes, I think I have to edit the Crush rule and modify: item_name or to be clear: I need to modify this in the decompiled crush map: root bmeta { id -4 # do not change unnecessarily id -254 class hdd # do not change unnecessarily id -256 class ssd # do not change unnecessarily # weight 0.000 alg straw2 hash 0 # rjenkins1 item gor-bmeta weight 0.000 root and item have to be modified to match the other host I moved the OSDs to, I think. What do you think? Best, Malte Am 18.10.23 um 11:30 schrieb Malte Stroem: > Hello Eugen, > > I was wrong. I am sorry. > > The PGs are not empty and orphaned. > > Most of the PGs are empty but a few are indeed used. > > And the pool for these PGs is still there. It is the metadata pool of > the erasure coded pool for RBDs. The cache tier pool was removed > successfully. > > So now we need to empty these OSDs. > > The device class was SSD. I changed it to HDD and moved the OSDs inside > the Crush tree to the other HDD OSDs of the host. > > I need to move the PGs away from the OSDs to other OSDs but I do not > know how to do it. > > Is using pg-upmap the solution? > > Is using the objectstore-tool the solution? > > Is moving the OSDs inside Crush to the right place the solution? > > Is migrating the metadata pool to another with another crush rule the > solution? > > The crush rule of this metadata pool looks like this: > > { >     "rule_id": 8, >     "rule_name": "rbd-meta", >     "ruleset": 6, >     "type": 1, >     "min_size": 1, >     "max_size": 10, >     "steps": [ >         { >             "op": "take", >             "item": -4, >             "item_name": "bmeta" >         }, >         { >             "op": "chooseleaf_firstn", >             "num": 0, >             "type": "host" >         }, >         { >             "op": "emit" >         } >     ] > } > > When stopping one of the OSDs the status gets degraded. > > How to move PGs away from the OSDs? > > How to let the pool use other OSDs? > > Changing the crush rule? > > Best, > Malte > > Am 05.10.23 um 11:35 schrieb Malte Stroem: >> Hello Eugen, Hello Joachim, >> >> @Joachim: Interesting! And you got empty PGs, too? How did you solve >> the problem? >> >> @Eugen: This is one of our biggest clusters and we're in the process >> to migrate from Nautilus to Octopus and to migrate from CentOS to Ubuntu. >> >> The cache tier pool's OSDs were still version 14 OSDs. Most of the >> other OSDs are version 15 already. >> >> So I tested the command: >> >> ceph-objectstore-tool --data-path /path/to/osd --op remove --pgid 3.0 >> --force >> >> in a test cluster environment and this worked fine. >> >> But the test scenario was not similar to our productive environment >> and the PG wasn't empty. >> >> I did not find a way to emulate the same situation in the test >> scenario, yet. >> >> Best, >> Malte >> >> Am 05.10.23 um 11:03 schrieb Eugen Block: >>> I know, I know... but since we are already using it (for years) I >>> have to check how to remove it safely, maybe as long as we're on >>> Pacific. ;-) >>> >>> Zitat von Joachim Kraftmayer - ceph ambassador >>> <joachim.kraftmayer(a)clyso.com>om>: >>> >>>> @Eugen >>>> >>>> We have seen the same problems 8 years ago. I can only recommend >>>> never to use cache tiering in production. >>>> At Cephalocon this was part of my talk and as far as I remember >>>> cache tiering will also disappear from ceph soon. >>>> >>>> Cache tiering has been deprecated in the Reef release as it has >>>> lacked a maintainer for a very long time. This does not mean it will >>>> be certainly removed, but we may choose to remove it without much >>>> further notice. >>>> >>>> https://docs.ceph.com/en/latest/rados/operations/cache-tiering/ >>>> >>>> Regards, Joachim >>>> >>>> >>>> ___________________________________ >>>> ceph ambassador DACH >>>> ceph consultant since 2012 >>>> >>>> Clyso GmbH - Premier Ceph Foundation Member >>>> >>>> https://www.clyso.com/ >>>> >>>> Am 05.10.23 um 10:02 schrieb Eugen Block: >>>>> Which ceph version is this? I'm trying to understand how removing a >>>>> pool leaves the PGs of that pool... Do you have any logs or >>>>> something from when you removed the pool? >>>>> We'll have to deal with a cache tier in the forseeable future as >>>>> well so this is quite relevant for us as well. Maybe I'll try to >>>>> reproduce it in a test cluster first. >>>>> Are those SSDs exclusively for the cache tier or are they used by >>>>> other pools as well? If they were used only for the cache tier you >>>>> should be able to just remove them without any risk. But as I said, >>>>> I'd rather try to understand before purging them. >>>>> >>>>> >>>>> Zitat von Malte Stroem <malte.stroem(a)gmail.com>om>: >>>>> >>>>>> Hello Eugen, >>>>>> >>>>>> yes, we followed the documentation and everything worked fine. The >>>>>> cache is gone. >>>>>> >>>>>> Removing the pool worked well. Everything is clean. >>>>>> >>>>>> The PGs are empty active+clean. >>>>>> >>>>>> Possible solutions: >>>>>> >>>>>> 1. >>>>>> >>>>>> ceph pg {pg-id} mark_unfound_lost delete >>>>>> >>>>>> I do not think this is the right way since it is for PGs with >>>>>> status unfound. But it could work also. >>>>>> >>>>>> 2. >>>>>> >>>>>> Set the following for the three disk: >>>>>> >>>>>> ceph osd lost {osd-id} >>>>>> >>>>>> I am not sure how the cluster will react to this. >>>>>> >>>>>> 3. >>>>>> >>>>>> ceph-objectstore-tool --data-path /path/to/osd --op remove --pgid >>>>>> 3.0 --force >>>>>> >>>>>> Now, will the cluster accept the removed PG status? >>>>>> >>>>>> 4. >>>>>> >>>>>> The three disks are still presented in the crush rule, class ssd, >>>>>> each single OSD under one host entry. >>>>>> >>>>>> What if I remove them from crush? >>>>>> >>>>>> Do you have a better idea, Eugen? >>>>>> >>>>>> Best, >>>>>> Malte >>>>>> >>>>>> Am 04.10.23 um 09:21 schrieb Eugen Block: >>>>>>> Hi, >>>>>>> >>>>>>> just for clarity, you're actually talking about the cache tier as >>>>>>> described in the docs [1]? And you followed the steps until 'ceph >>>>>>> osd tier remove cold-storage hot-storage' successfully? And the >>>>>>> pool has been really deleted successfully ('ceph osd pool ls >>>>>>> detail')? >>>>>>> >>>>>>> [1] >>>>>>> https://docs.ceph.com/en/latest/rados/operations/cache-tiering/#removing-a-… >>>>>>> >>>>>>> Zitat von Malte Stroem <malte.stroem(a)gmail.com>om>: >>>>>>> >>>>>>>> Hello, >>>>>>>> >>>>>>>> we removed an SSD cache tier and its pool. >>>>>>>> >>>>>>>> The PGs for the pool do still exist. >>>>>>>> >>>>>>>> The cluster is healthy. >>>>>>>> >>>>>>>> The PGs are empty and they reside on the cache tier pool's SSDs. >>>>>>>> >>>>>>>> We like to take out the disks but it is not possible. The >>>>>>>> cluster sees the PGs and answers with a HEALTH_WARN. >>>>>>>> >>>>>>>> Because of the replication of three there are still 128 PGs on >>>>>>>> three of the 24 OSDs. We were able to remove the other OSDs. >>>>>>>> >>>>>>>> Summary: >>>>>>>> >>>>>>>> - pool removed >>>>>>>> - 3 x 128 empty PGs still exist >>>>>>>> - 3 of 24 OSDs still exist >>>>>>>> >>>>>>>> How is it possible to remove these empty and healthy PGs? >>>>>>>> >>>>>>>> The only way I found was something like: >>>>>>>> >>>>>>>> ceph pg {pg-id} mark_unfound_lost delete >>>>>>>> >>>>>>>> Is that the right way? >>>>>>>> >>>>>>>> Some output of: >>>>>>>> >>>>>>>> ceph pg ls-by-osd 23 >>>>>>>> >>>>>>>> PG      OBJECTS  DEGRADED  MISPLACED  UNFOUND  BYTES OMAP_BYTES* >>>>>>>> OMAP_KEYS*  LOG   STATE         SINCE VERSION         REPORTED >>>>>>>> UP ACTING             SCRUB_STAMP DEEP_SCRUB_STAMP >>>>>>>> 3.0          0         0          0        0 0            0   0 >>>>>>>>     0  active+clean    27h 0'0    2627265:196316 [15,6,23]p15 >>>>>>>> [15,6,23]p15 2023-09-28T12:41:52.982955+0200 >>>>>>>> 2023-09-27T06:48:23.265838+0200 >>>>>>>> 3.1          0         0          0        0 0            0   0 >>>>>>>>     0  active+clean     9h 0'0    2627266:19330 [6,23,15]p6 >>>>>>>> [6,23,15]p6 2023-09-29T06:30:57.630016+0200 >>>>>>>> 2023-09-27T22:58:21.992451+0200 >>>>>>>> 3.2          0         0          0        0 0            0   0 >>>>>>>>     0  active+clean     2h 0'0   2627265:1135185 [23,15,6]p23 >>>>>>>> [23,15,6]p23 2023-09-29T13:42:07.346658+0200 >>>>>>>> 2023-09-24T14:31:52.844427+0200 >>>>>>>> 3.3          0         0          0        0 0            0   0 >>>>>>>>     0  active+clean    13h 0'0    2627266:193170 [6,15,23]p6 >>>>>>>> [6,15,23]p6 2023-09-29T01:56:54.517337+0200 >>>>>>>> 2023-09-27T17:47:24.961279+0200 >>>>>>>> 3.4          0         0          0        0 0            0   0 >>>>>>>>     0  active+clean    14h 0'0   2627265:2343551 [23,6,15]p23 >>>>>>>> [23,6,15]p23 2023-09-29T00:47:47.548860+0200 >>>>>>>> 2023-09-25T09:39:51.259304+0200 >>>>>>>> 3.5          0         0          0        0 0            0   0 >>>>>>>>     0  active+clean     2h 0'0    2627265:194111 [15,6,23]p15 >>>>>>>> [15,6,23]p15 2023-09-29T13:28:48.879959+0200 >>>>>>>> 2023-09-26T15:35:44.217302+0200 >>>>>>>> 3.6          0         0          0        0 0            0   0 >>>>>>>>     0  active+clean     6h 0'0   2627265:2345717 [23,15,6]p23 >>>>>>>> [23,15,6]p23 2023-09-29T09:26:02.534825+0200 >>>>>>>> 2023-09-27T21:56:57.500126+0200 >>>>>>>> >>>>>>>> Best regards, >>>>>>>> Malte >>>>>>>> _______________________________________________ >>>>>>>> ceph-users mailing list -- ceph-users(a)ceph.io >>>>>>>> To unsubscribe send an email to ceph-users-leave(a)ceph.io >>>>>>> >>>>>>> >>>>>>> _______________________________________________ >>>>>>> ceph-users mailing list -- ceph-users(a)ceph.io >>>>>>> To unsubscribe send an email to ceph-users-leave(a)ceph.io >>>>> >>>>> >>>>> _______________________________________________ >>>>> ceph-users mailing list -- ceph-users(a)ceph.io >>>>> To unsubscribe send an email to ceph-users-leave(a)ceph.io >>> >>> >>>

Hello Eugen, before answering your questions: Creating a new simple replicated crush rule and setting it to the metadata pool was the solution. Yes, the cache tier pool and the metadata pool for the RBDs. Yes. the same SSDs. No, it was located on the SSDs from the beginning. Looking at the device classes was wrong. Just setting the right crush rule was the right way. Thank you very much and please feel free to ask me about the cache tier process or if you need some help. Best regards, Malte On 18.10.23 11:50, Eugen Block wrote: > Hi, > >> So now we need to empty these OSDs. >> >> The device class was SSD. I changed it to HDD and moved the OSDs >> inside the Crush tree to the other HDD OSDs of the host. >> I need to move the PGs away from the OSDs to other OSDs but I do not >> know how to do it. > > your crush rule doesn't specify a device class so moving them around > doesn't really help (as you already noticed). Are other pools using that > crush rule? You can see the applied rule IDs in 'ceph osd pool ls > detail' output. If other pools use the same rule, make sure the cluster > can handle data movement if you change it. Test the modified rule with > crushtool first, and maybe in your test cluster as well. > If you add a device class statement (step take default class hdd) to the > rule the SSDs will be drained automatically, at least they should. > > But before you change anyhting I just want to make sure I understand > correctly: > > - Your cache tier was on SSDs which need to be removed. > - Cache tier was removed successfully. > - But since the rbd-meta pool is not aware of device classes it used the > same SSDs. > > Don't change the root bmeta, only the crush rule "rbd-meta", here's an > example from a replicated pool: > > # rules > rule replicated_ruleset { >         id 0 >         type replicated >         min_size 1 >         max_size 10 >         step take default class hdd >         step chooseleaf firstn 0 type host >         step emit > } > > > Zitat von Malte Stroem <malte.stroem(a)gmail.com>om>: > >> Hello Eugen, >> >> I was wrong. I am sorry. >> >> The PGs are not empty and orphaned. >> >> Most of the PGs are empty but a few are indeed used. >> >> And the pool for these PGs is still there. It is the metadata pool of >> the erasure coded pool for RBDs. The cache tier pool was removed >> successfully. >> >> So now we need to empty these OSDs. >> >> The device class was SSD. I changed it to HDD and moved the OSDs >> inside the Crush tree to the other HDD OSDs of the host. >> >> I need to move the PGs away from the OSDs to other OSDs but I do not >> know how to do it. >> >> Is using pg-upmap the solution? >> >> Is using the objectstore-tool the solution? >> >> Is moving the OSDs inside Crush to the right place the solution? >> >> Is migrating the metadata pool to another with another crush rule the >> solution? >> >> The crush rule of this metadata pool looks like this: >> >> { >>     "rule_id": 8, >>     "rule_name": "rbd-meta", >>     "ruleset": 6, >>     "type": 1, >>     "min_size": 1, >>     "max_size": 10, >>     "steps": [ >>         { >>             "op": "take", >>             "item": -4, >>             "item_name": "bmeta" >>         }, >>         { >>             "op": "chooseleaf_firstn", >>             "num": 0, >>             "type": "host" >>         }, >>         { >>             "op": "emit" >>         } >>     ] >> } >> >> When stopping one of the OSDs the status gets degraded. >> >> How to move PGs away from the OSDs? >> >> How to let the pool use other OSDs? >> >> Changing the crush rule? >> >> Best, >> Malte >> >> Am 05.10.23 um 11:35 schrieb Malte Stroem: >>> Hello Eugen, Hello Joachim, >>> >>> @Joachim: Interesting! And you got empty PGs, too? How did you solve >>> the problem? >>> >>> @Eugen: This is one of our biggest clusters and we're in the process >>> to migrate from Nautilus to Octopus and to migrate from CentOS to >>> Ubuntu. >>> >>> The cache tier pool's OSDs were still version 14 OSDs. Most of the >>> other OSDs are version 15 already. >>> >>> So I tested the command: >>> >>> ceph-objectstore-tool --data-path /path/to/osd --op remove --pgid 3.0 >>> --force >>> >>> in a test cluster environment and this worked fine. >>> >>> But the test scenario was not similar to our productive environment >>> and the PG wasn't empty. >>> >>> I did not find a way to emulate the same situation in the test >>> scenario, yet. >>> >>> Best, >>> Malte >>> >>> Am 05.10.23 um 11:03 schrieb Eugen Block: >>>> I know, I know... but since we are already using it (for years) I >>>> have to check how to remove it safely, maybe as long as we're on >>>> Pacific. ;-) >>>> >>>> Zitat von Joachim Kraftmayer - ceph ambassador >>>> <joachim.kraftmayer(a)clyso.com>om>: >>>> >>>>> @Eugen >>>>> >>>>> We have seen the same problems 8 years ago. I can only recommend >>>>> never to use cache tiering in production. >>>>> At Cephalocon this was part of my talk and as far as I remember >>>>> cache tiering will also disappear from ceph soon. >>>>> >>>>> Cache tiering has been deprecated in the Reef release as it has >>>>> lacked a maintainer for a very long time. This does not mean it >>>>> will be certainly removed, but we may choose to remove it without >>>>> much further notice. >>>>> >>>>> https://docs.ceph.com/en/latest/rados/operations/cache-tiering/ >>>>> >>>>> Regards, Joachim >>>>> >>>>> >>>>> ___________________________________ >>>>> ceph ambassador DACH >>>>> ceph consultant since 2012 >>>>> >>>>> Clyso GmbH - Premier Ceph Foundation Member >>>>> >>>>> https://www.clyso.com/ >>>>> >>>>> Am 05.10.23 um 10:02 schrieb Eugen Block: >>>>>> Which ceph version is this? I'm trying to understand how removing >>>>>> a pool leaves the PGs of that pool... Do you have any logs or >>>>>> something from when you removed the pool? >>>>>> We'll have to deal with a cache tier in the forseeable future as >>>>>> well so this is quite relevant for us as well. Maybe I'll try to >>>>>> reproduce it in a test cluster first. >>>>>> Are those SSDs exclusively for the cache tier or are they used by >>>>>> other pools as well? If they were used only for the cache tier you >>>>>> should be able to just remove them without any risk. But as I >>>>>> said, I'd rather try to understand before purging them. >>>>>> >>>>>> >>>>>> Zitat von Malte Stroem <malte.stroem(a)gmail.com>om>: >>>>>> >>>>>>> Hello Eugen, >>>>>>> >>>>>>> yes, we followed the documentation and everything worked fine. >>>>>>> The cache is gone. >>>>>>> >>>>>>> Removing the pool worked well. Everything is clean. >>>>>>> >>>>>>> The PGs are empty active+clean. >>>>>>> >>>>>>> Possible solutions: >>>>>>> >>>>>>> 1. >>>>>>> >>>>>>> ceph pg {pg-id} mark_unfound_lost delete >>>>>>> >>>>>>> I do not think this is the right way since it is for PGs with >>>>>>> status unfound. But it could work also. >>>>>>> >>>>>>> 2. >>>>>>> >>>>>>> Set the following for the three disk: >>>>>>> >>>>>>> ceph osd lost {osd-id} >>>>>>> >>>>>>> I am not sure how the cluster will react to this. >>>>>>> >>>>>>> 3. >>>>>>> >>>>>>> ceph-objectstore-tool --data-path /path/to/osd --op remove --pgid >>>>>>> 3.0 --force >>>>>>> >>>>>>> Now, will the cluster accept the removed PG status? >>>>>>> >>>>>>> 4. >>>>>>> >>>>>>> The three disks are still presented in the crush rule, class ssd, >>>>>>> each single OSD under one host entry. >>>>>>> >>>>>>> What if I remove them from crush? >>>>>>> >>>>>>> Do you have a better idea, Eugen? >>>>>>> >>>>>>> Best, >>>>>>> Malte >>>>>>> >>>>>>> Am 04.10.23 um 09:21 schrieb Eugen Block: >>>>>>>> Hi, >>>>>>>> >>>>>>>> just for clarity, you're actually talking about the cache tier >>>>>>>> as described in the docs [1]? And you followed the steps until >>>>>>>> 'ceph osd tier remove cold-storage hot-storage' successfully? >>>>>>>> And the pool has been really deleted successfully ('ceph osd >>>>>>>> pool ls detail')? >>>>>>>> >>>>>>>> [1] >>>>>>>> https://docs.ceph.com/en/latest/rados/operations/cache-tiering/#removing-a-… >>>>>>>> >>>>>>>> Zitat von Malte Stroem <malte.stroem(a)gmail.com>om>: >>>>>>>> >>>>>>>>> Hello, >>>>>>>>> >>>>>>>>> we removed an SSD cache tier and its pool. >>>>>>>>> >>>>>>>>> The PGs for the pool do still exist. >>>>>>>>> >>>>>>>>> The cluster is healthy. >>>>>>>>> >>>>>>>>> The PGs are empty and they reside on the cache tier pool's SSDs. >>>>>>>>> >>>>>>>>> We like to take out the disks but it is not possible. The >>>>>>>>> cluster sees the PGs and answers with a HEALTH_WARN. >>>>>>>>> >>>>>>>>> Because of the replication of three there are still 128 PGs on >>>>>>>>> three of the 24 OSDs. We were able to remove the other OSDs. >>>>>>>>> >>>>>>>>> Summary: >>>>>>>>> >>>>>>>>> - pool removed >>>>>>>>> - 3 x 128 empty PGs still exist >>>>>>>>> - 3 of 24 OSDs still exist >>>>>>>>> >>>>>>>>> How is it possible to remove these empty and healthy PGs? >>>>>>>>> >>>>>>>>> The only way I found was something like: >>>>>>>>> >>>>>>>>> ceph pg {pg-id} mark_unfound_lost delete >>>>>>>>> >>>>>>>>> Is that the right way? >>>>>>>>> >>>>>>>>> Some output of: >>>>>>>>> >>>>>>>>> ceph pg ls-by-osd 23 >>>>>>>>> >>>>>>>>> PG      OBJECTS  DEGRADED  MISPLACED  UNFOUND  BYTES >>>>>>>>> OMAP_BYTES* OMAP_KEYS*  LOG   STATE         SINCE >>>>>>>>> VERSION         REPORTED UP ACTING             SCRUB_STAMP >>>>>>>>> DEEP_SCRUB_STAMP >>>>>>>>> 3.0          0         0          0        0 0            0   0 >>>>>>>>>     0  active+clean    27h 0'0    2627265:196316 [15,6,23]p15 >>>>>>>>> [15,6,23]p15 2023-09-28T12:41:52.982955+0200 >>>>>>>>> 2023-09-27T06:48:23.265838+0200 >>>>>>>>> 3.1          0         0          0        0 0            0   0 >>>>>>>>>     0  active+clean     9h 0'0    2627266:19330 [6,23,15]p6 >>>>>>>>> [6,23,15]p6 2023-09-29T06:30:57.630016+0200 >>>>>>>>> 2023-09-27T22:58:21.992451+0200 >>>>>>>>> 3.2          0         0          0        0 0            0   0 >>>>>>>>>     0  active+clean     2h 0'0   2627265:1135185 [23,15,6]p23 >>>>>>>>> [23,15,6]p23 2023-09-29T13:42:07.346658+0200 >>>>>>>>> 2023-09-24T14:31:52.844427+0200 >>>>>>>>> 3.3          0         0          0        0 0            0   0 >>>>>>>>>     0  active+clean    13h 0'0    2627266:193170 [6,15,23]p6 >>>>>>>>> [6,15,23]p6 2023-09-29T01:56:54.517337+0200 >>>>>>>>> 2023-09-27T17:47:24.961279+0200 >>>>>>>>> 3.4          0         0          0        0 0            0   0 >>>>>>>>>     0  active+clean    14h 0'0   2627265:2343551 [23,6,15]p23 >>>>>>>>> [23,6,15]p23 2023-09-29T00:47:47.548860+0200 >>>>>>>>> 2023-09-25T09:39:51.259304+0200 >>>>>>>>> 3.5          0         0          0        0 0            0   0 >>>>>>>>>     0  active+clean     2h 0'0    2627265:194111 [15,6,23]p15 >>>>>>>>> [15,6,23]p15 2023-09-29T13:28:48.879959+0200 >>>>>>>>> 2023-09-26T15:35:44.217302+0200 >>>>>>>>> 3.6          0         0          0        0 0            0   0 >>>>>>>>>     0  active+clean     6h 0'0   2627265:2345717 [23,15,6]p23 >>>>>>>>> [23,15,6]p23 2023-09-29T09:26:02.534825+0200 >>>>>>>>> 2023-09-27T21:56:57.500126+0200 >>>>>>>>> >>>>>>>>> Best regards, >>>>>>>>> Malte >>>>>>>>> _______________________________________________ >>>>>>>>> ceph-users mailing list -- ceph-users(a)ceph.io >>>>>>>>> To unsubscribe send an email to ceph-users-leave(a)ceph.io >>>>>>>> >>>>>>>> >>>>>>>> _______________________________________________ >>>>>>>> ceph-users mailing list -- ceph-users(a)ceph.io >>>>>>>> To unsubscribe send an email to ceph-users-leave(a)ceph.io >>>>>> >>>>>> >>>>>> _______________________________________________ >>>>>> ceph-users mailing list -- ceph-users(a)ceph.io >>>>>> To unsubscribe send an email to ceph-users-leave(a)ceph.io >>>> >>>> >>>> > > >