Re: PGs inconsistent, do I fear data loss?

[Denes Dolhay <denke@xxxxxxxxxxxx>]

Hello,

I have a trick question for Mr. Turner's scenario:
Let's assume size=2, min_size=1
-We are looking at pg "A" acting [1, 2]
-osd 1 goes down, OK
-osd 1 comes back up, backfill of pg "A" commences from osd 2 to osd 1, OK
-osd 2 goes down (and therefore pg "A" 's backfill to osd 1 is incomplete and stopped) not OK, but this is the case...
--> In this event, why does osd 1 accept IO to pg "A" knowing full well, that it's data is outdated and will cause an inconsistent state?
Wouldn't it be prudent to deny io to pg "A" until either
-osd 2 comes back (therefore we have a clean osd in the acting group) ... backfill would continue to osd 1 of course
-or data in pg "A" is manually marked as lost, and then continues operation from osd 1 's (outdated) copy?

Thanks in advance, I'm really curious!
Denes.

On 11/01/2017 06:33 PM, Mario Giammarco wrote:

I have read your post then read the thread you suggested, very interesting.

Then I read again your post and understood better.

The most important thing is that even with min_size=1 writes are acknowledged after ceph wrote size=2 copies.

In the thread above there is: 

As David already said, when all OSDs are up and in for a PG Ceph will wait for ALL OSDs to Ack the write. Writes in RADOS are always synchronous.

Only when OSDs go down you need at least min_size OSDs up before writes or reads are accepted.

So if min_size = 2 and size = 3 you need at least 2 OSDs online for I/O to take place.

You then show me a sequence of events that may happen in some use cases.

I tell you my use case which is quite different. We use ceph under proxmox. The servers have disks on raid 5 (I agree that it is better to expose single disks to Ceph but it is late). 

So it is unlikely that a ceph disk fails because of raid. If a disks fail probabably is because the entire server has failed (and we need to provide business availability in this case) and so it will never come up again so in my situation your sequence of events will never happen.

What shocked me is that I did not expect to see so many inconsistencies.

Thanks,

Mario

2017-11-01 16:45 GMT+01:00 David Turner <drakonstein@xxxxxxxxx>:

It looks like you're running with a size = 2 and min_size = 1 (the min_size is a guess, the size is based on how many osds belong to your problem PGs).  Here's some good reading for you.  https://www.spinics.net/lists/ceph-users/msg32895.html

Basically the jist is that when running with size = 2 you should assume that data loss is an eventuality and choose that it is ok for your use case.  This can be mitigated by using min_size = 2, but then your pool will block while an OSD is down and you'll have to manually go in and change the min_size temporarily to perform maintenance.

All it takes for data loss is that an osd on server 1 is marked down and a write happens to an osd on server 2.  Now the osd on server 2 goes down before the osd on server 1 has finished backfilling and the first osd receives a request to modify data in the object that it doesn't know the current state of.  Tada, you have data loss.

How likely is this to happen... eventually it will.  PG subfolder splitting (if you're using filestore) will occasionally take long enough to perform the task that the osd is marked down while it's still running, and this usually happens for some time all over the cluster when it does.  Another option is something that causes segfaults in the osds; another is restarting a node before all pgs are done backfilling/recovering; OOM killer; power outages; etc; etc.

Why does min_size = 2 prevent this?  Because for a write to be acknowledged by the cluster, it has to be written to every OSD that is up as long as there are at least min_size available.  This means that every write is acknowledged by at least 2 osds every time.  If you're running with size = 2, then both copies of the data need to be online for a write to happen and thus can never have a write that the other does not.  If you're running with size = 3, then you always have a majority of the OSDs online receiving a write and they can both agree on the correct data to give to the third when it comes back up.

On Wed, Nov 1, 2017 at 3:31 AM Mario Giammarco <mgiammarco@xxxxxxxxx> wrote:

Sure here it is ceph -s:

cluster:
   id:     8bc45d9a-ef50-4038-8e1b-1f25ac46c945
   health: HEALTH_ERR
           100 scrub errors
           Possible data damage: 56 pgs inconsistent
 
 services:
   mon: 3 daemons, quorum 0,1,pve3
   mgr: pve3(active)
   osd: 3 osds: 3 up, 3 in
 
 data:
   pools:   1 pools, 256 pgs
   objects: 269k objects, 1007 GB
   usage:   2050 GB used, 1386 GB / 3436 GB avail
   pgs:     200 active+clean
            56  active+clean+inconsistent 

---

ceph health detail :

PG_DAMAGED Possible data damage: 56 pgs inconsistent
   pg 2.6 is active+clean+inconsistent, acting [1,0]
   pg 2.19 is active+clean+inconsistent, acting [1,2]
   pg 2.1e is active+clean+inconsistent, acting [1,2]
   pg 2.1f is active+clean+inconsistent, acting [1,2]
   pg 2.24 is active+clean+inconsistent, acting [0,2]
   pg 2.25 is active+clean+inconsistent, acting [2,0]
   pg 2.36 is active+clean+inconsistent, acting [1,0]
   pg 2.3d is active+clean+inconsistent, acting [1,2]
   pg 2.4b is active+clean+inconsistent, acting [1,0]
   pg 2.4c is active+clean+inconsistent, acting [0,2]
   pg 2.4d is active+clean+inconsistent, acting [1,2]
   pg 2.4f is active+clean+inconsistent, acting [1,2]
   pg 2.50 is active+clean+inconsistent, acting [1,2]
   pg 2.52 is active+clean+inconsistent, acting [1,2]
   pg 2.56 is active+clean+inconsistent, acting [1,0]
   pg 2.5b is active+clean+inconsistent, acting [1,2]
   pg 2.5c is active+clean+inconsistent, acting [1,2]
   pg 2.5d is active+clean+inconsistent, acting [1,0]
   pg 2.5f is active+clean+inconsistent, acting [1,2]
   pg 2.71 is active+clean+inconsistent, acting [0,2]
   pg 2.75 is active+clean+inconsistent, acting [1,2]
   pg 2.77 is active+clean+inconsistent, acting [1,2]
   pg 2.79 is active+clean+inconsistent, acting [1,2]
   pg 2.7e is active+clean+inconsistent, acting [1,2]
   pg 2.83 is active+clean+inconsistent, acting [1,0]
   pg 2.8a is active+clean+inconsistent, acting [1,0]
   pg 2.92 is active+clean+inconsistent, acting [1,2]
   pg 2.98 is active+clean+inconsistent, acting [1,0]
   pg 2.9a is active+clean+inconsistent, acting [1,0]
   pg 2.9e is active+clean+inconsistent, acting [1,0]
   pg 2.9f is active+clean+inconsistent, acting [1,2]
   pg 2.c6 is active+clean+inconsistent, acting [0,2]
   pg 2.c7 is active+clean+inconsistent, acting [1,0]
   pg 2.c8 is active+clean+inconsistent, acting [1,2]
   pg 2.cb is active+clean+inconsistent, acting [1,2]
   pg 2.cd is active+clean+inconsistent, acting [1,2]
   pg 2.ce is active+clean+inconsistent, acting [1,2]
   pg 2.d2 is active+clean+inconsistent, acting [2,1]
   pg 2.da is active+clean+inconsistent, acting [1,0]
   pg 2.de is active+clean+inconsistent, acting [1,2]
   pg 2.e1 is active+clean+inconsistent, acting [1,2]
   pg 2.e4 is active+clean+inconsistent, acting [1,0]
   pg 2.e6 is active+clean+inconsistent, acting [0,2]
   pg 2.e8 is active+clean+inconsistent, acting [1,2]
   pg 2.ee is active+clean+inconsistent, acting [1,0]
   pg 2.f9 is active+clean+inconsistent, acting [1,2]
   pg 2.fa is active+clean+inconsistent, acting [1,0]
   pg 2.fb is active+clean+inconsistent, acting [1,2]
   pg 2.fc is active+clean+inconsistent, acting [1,2]
   pg 2.fe is active+clean+inconsistent, acting [1,0]
   pg 2.ff is active+clean+inconsistent, acting [1,0]

and ceph pg 2.6 query: 

{
   "state": "active+clean+inconsistent",
   "snap_trimq": "[]",
   "epoch": 1513,
   "up": [
       1,
       0
   ],
   "acting": [
       1,
       0
   ],
   "actingbackfill": [
       "0",
       "1"
   ],
   "info": {
       "pgid": "2.6",
       "last_update": "1513'89145",
       "last_complete": "1513'89145",
       "log_tail": "1503'87586",
       "last_user_version": 330583,
       "last_backfill": "MAX",
       "last_backfill_bitwise": 0,
       "purged_snaps": [
           {
               "start": "1",
               "length": "178"
           },
           {
               "start": "17a",
               "length": "3d"
           },
           {
               "start": "1b8",
               "length": "1"
           },
           {
               "start": "1ba",
               "length": "1"
           },
           {
               "start": "1bc",
               "length": "1"
           },
           {
               "start": "1be",
               "length": "44"
           },
           {
               "start": "205",
               "length": "12c"
           },
           {
               "start": "332",
               "length": "1"
           },
           {
               "start": "334",
               "length": "1"
           },
           {
               "start": "336",
               "length": "1"
           },
           {
               "start": "338",
               "length": "1"
           },
           {
               "start": "33a",
               "length": "1"
           }
       ],
       "history": {
           "epoch_created": 90,
           "epoch_pool_created": 90,
           "last_epoch_started": 1339,
           "last_interval_started": 1338,
           "last_epoch_clean": 1339,
           "last_interval_clean": 1338,
           "last_epoch_split": 0,
           "last_epoch_marked_full": 0,
           "same_up_since": 1338,
           "same_interval_since": 1338,
           "same_primary_since": 1338,
           "last_scrub": "1513'89112",
           "last_scrub_stamp": "2017-11-01 05:52:21.259654",
           "last_deep_scrub": "1513'89112",
           "last_deep_scrub_stamp": "2017-11-01 05:52:21.259654",
           "last_clean_scrub_stamp": "2017-10-25 04:25:09.830840"
       },
       "stats": {
           "version": "1513'89145",
           "reported_seq": "422820",
           "reported_epoch": "1513",
           "state": "active+clean+inconsistent",
           "last_fresh": "2017-11-01 08:11:38.411784",
           "last_change": "2017-11-01 05:52:21.259789",
           "last_active": "2017-11-01 08:11:38.411784",
           "last_peered": "2017-11-01 08:11:38.411784",
           "last_clean": "2017-11-01 08:11:38.411784",
           "last_became_active": "2017-10-15 20:36:33.644567",
           "last_became_peered": "2017-10-15 20:36:33.644567",
           "last_unstale": "2017-11-01 08:11:38.411784",
           "last_undegraded": "2017-11-01 08:11:38.411784",
           "last_fullsized": "2017-11-01 08:11:38.411784",
           "mapping_epoch": 1338,
           "log_start": "1503'87586",
           "ondisk_log_start": "1503'87586",
           "created": 90,
           "last_epoch_clean": 1339,
           "parent": "0.0",
           "parent_split_bits": 0,
           "last_scrub": "1513'89112",
           "last_scrub_stamp": "2017-11-01 05:52:21.259654",
           "last_deep_scrub": "1513'89112",
           "last_deep_scrub_stamp": "2017-11-01 05:52:21.259654",
           "last_clean_scrub_stamp": "2017-10-25 04:25:09.830840",
           "log_size": 1559,
           "ondisk_log_size": 1559,
           "stats_invalid": false,
           "dirty_stats_invalid": false,
           "omap_stats_invalid": false,
           "hitset_stats_invalid": false,
           "hitset_bytes_stats_invalid": false,
           "pin_stats_invalid": false,
           "stat_sum": {
               "num_bytes": 3747886080,
               "num_objects": 958,
               "num_object_clones": 295,
               "num_object_copies": 1916,
               "num_objects_missing_on_primary": 0,
               "num_objects_missing": 0,
               "num_objects_degraded": 0,
               "num_objects_misplaced": 0,
               "num_objects_unfound": 0,
               "num_objects_dirty": 958,
               "num_whiteouts": 0,
               "num_read": 333428,
               "num_read_kb": 135550185,
               "num_write": 79221,
               "num_write_kb": 13441239,
               "num_scrub_errors": 1,
               "num_shallow_scrub_errors": 0,
               "num_deep_scrub_errors": 1,
               "num_objects_recovered": 245,
               "num_bytes_recovered": 1012833792,
               "num_keys_recovered": 6,
               "num_objects_omap": 0,
               "num_objects_hit_set_archive": 0,
               "num_bytes_hit_set_archive": 0,
               "num_flush": 0,
               "num_flush_kb": 0,
               "num_evict": 0,
               "num_evict_kb": 0,
               "num_promote": 0,
               "num_flush_mode_high": 0,
               "num_flush_mode_low": 0,
               "num_evict_mode_some": 0,
               "num_evict_mode_full": 0,
               "num_objects_pinned": 0,
               "num_legacy_snapsets": 0
           },
           "up": [
               1,
               0
           ],
           "acting": [
               1,
               0
           ],
           "blocked_by": [],
           "up_primary": 1,
           "acting_primary": 1
       },
       "empty": 0,
       "dne": 0,
       "incomplete": 0,
       "last_epoch_started": 1339,
       "hit_set_history": {
           "current_last_update": "0'0",
           "history": []
       }
   },
   "peer_info": [
       {
           "peer": "0",
           "pgid": "2.6",
           "last_update": "1513'89145",
           "last_complete": "1513'89145",
           "log_tail": "1274'68440",
           "last_user_version": 315687,
           "last_backfill": "MAX",
           "last_backfill_bitwise": 0,
           "purged_snaps": [
               {
                   "start": "1",
                   "length": "178"
               },
               {
                   "start": "17a",
                   "length": "3d"
               },
               {
                   "start": "1b8",
                   "length": "1"
               },
               {
                   "start": "1ba",
                   "length": "1"
               },
               {
                   "start": "1bc",
                   "length": "1"
               },
               {
                   "start": "1be",
                   "length": "44"
               },
               {
                   "start": "205",
                   "length": "82"
               },
               {
                   "start": "288",
                   "length": "1"
               },
               {
                   "start": "28a",
                   "length": "1"
               },
               {
                   "start": "28c",
                   "length": "1"
               },
               {
                   "start": "28e",
                   "length": "1"
               },
               {
                   "start": "290",
                   "length": "1"
               }
           ],
           "history": {
               "epoch_created": 90,
               "epoch_pool_created": 90,
               "last_epoch_started": 1339,
               "last_interval_started": 1338,
               "last_epoch_clean": 1339,
               "last_interval_clean": 1338,
               "last_epoch_split": 0,
               "last_epoch_marked_full": 0,
               "same_up_since": 1338,
               "same_interval_since": 1338,
               "same_primary_since": 1338,
               "last_scrub": "1513'89112",
               "last_scrub_stamp": "2017-11-01 05:52:21.259654",
               "last_deep_scrub": "1513'89112",
               "last_deep_scrub_stamp": "2017-11-01 05:52:21.259654",
               "last_clean_scrub_stamp": "2017-10-25 04:25:09.830840"
           },
           "stats": {
               "version": "1337'71465",
               "reported_seq": "347015",
               "reported_epoch": "1338",
               "state": "active+undersized+degraded",
               "last_fresh": "2017-10-15 20:35:36.930611",
               "last_change": "2017-10-15 20:30:35.752042",
               "last_active": "2017-10-15 20:35:36.930611",
               "last_peered": "2017-10-15 20:35:36.930611",
               "last_clean": "2017-10-15 20:30:01.443288",
               "last_became_active": "2017-10-15 20:30:35.752042",
               "last_became_peered": "2017-10-15 20:30:35.752042",
               "last_unstale": "2017-10-15 20:35:36.930611",
               "last_undegraded": "2017-10-15 20:30:35.749043",
               "last_fullsized": "2017-10-15 20:30:35.749043",
               "mapping_epoch": 1338,
               "log_start": "1274'68440",
               "ondisk_log_start": "1274'68440",
               "created": 90,
               "last_epoch_clean": 1331,
               "parent": "0.0",
               "parent_split_bits": 0,
               "last_scrub": "1294'71370",
               "last_scrub_stamp": "2017-10-15 09:27:31.756027",
               "last_deep_scrub": "1284'70813",
               "last_deep_scrub_stamp": "2017-10-14 06:35:57.556773",
               "last_clean_scrub_stamp": "2017-10-15 09:27:31.756027",
               "log_size": 3025,
               "ondisk_log_size": 3025,
               "stats_invalid": false,
               "dirty_stats_invalid": false,
               "omap_stats_invalid": false,
               "hitset_stats_invalid": false,
               "hitset_bytes_stats_invalid": false,
               "pin_stats_invalid": false,
               "stat_sum": {
                   "num_bytes": 3555027456,
                   "num_objects": 917,
                   "num_object_clones": 255,
                   "num_object_copies": 1834,
                   "num_objects_missing_on_primary": 0,
                   "num_objects_missing": 0,
                   "num_objects_degraded": 917,
                   "num_objects_misplaced": 0,
                   "num_objects_unfound": 0,
                   "num_objects_dirty": 917,
                   "num_whiteouts": 0,
                   "num_read": 275095,
                   "num_read_kb": 111713846,
                   "num_write": 64324,
                   "num_write_kb": 11365374,
                   "num_scrub_errors": 0,
                   "num_shallow_scrub_errors": 0,
                   "num_deep_scrub_errors": 0,
                   "num_objects_recovered": 243,
                   "num_bytes_recovered": 1008594432,
                   "num_keys_recovered": 6,
                   "num_objects_omap": 0,
                   "num_objects_hit_set_archive": 0,
                   "num_bytes_hit_set_archive": 0,
                   "num_flush": 0,
                   "num_flush_kb": 0,
                   "num_evict": 0,
                   "num_evict_kb": 0,
                   "num_promote": 0,
                   "num_flush_mode_high": 0,
                   "num_flush_mode_low": 0,
                   "num_evict_mode_some": 0,
                   "num_evict_mode_full": 0,
                   "num_objects_pinned": 0,
                   "num_legacy_snapsets": 0
               },
               "up": [
                   1,
                   0
               ],
               "acting": [
                   1,
                   0
               ],
               "blocked_by": [],
               "up_primary": 1,
               "acting_primary": 1
           },
           "empty": 0,
           "dne": 0,
           "incomplete": 0,
           "last_epoch_started": 1339,
           "hit_set_history": {
               "current_last_update": "0'0",
               "history": []
           }
       }
   ],
   "recovery_state": [
       {
           "name": "Started/Primary/Active",
           "enter_time": "2017-10-15 20:36:33.574915",
           "might_have_unfound": [
               {
                   "osd": "0",
                   "status": "already probed"
               }
           ],
           "recovery_progress": {
               "backfill_targets": [],
               "waiting_on_backfill": [],
               "last_backfill_started": "MIN",
               "backfill_info": {
                   "begin": "MIN",
                   "end": "MIN",
                   "objects": []
               },
               "peer_backfill_info": [],
               "backfills_in_flight": [],
               "recovering": [],
               "pg_backend": {
                   "pull_from_peer": [],
                   "pushing": []
               }
           },
           "scrub": {
               "scrubber.epoch_start": "1338",
               "scrubber.active": false,
               "scrubber.state": "INACTIVE",
               "scrubber.start": "MIN",
               "scrubber.end": "MIN",
               "scrubber.subset_last_update": "0'0",
               "scrubber.deep": false,
               "scrubber.seed": 0,
               "scrubber.waiting_on": 0,
               "scrubber.waiting_on_whom": []
           }
       },
       {
           "name": "Started",
           "enter_time": "2017-10-15 20:36:32.592892"
       }
   ],
   "agent_state": {}
}

2017-10-30 23:30 GMT+01:00 Gregory Farnum <gfarnum@xxxxxxxxxx>:

You'll need to tell us exactly what error messages you're seeing, what the output of ceph -s is, and the output of pg query for the relevant PGs.
There's not a lot of documentation because much of this tooling is new, it's changing quickly, and most people don't have the kinds of problems that turn out to be unrepairable. We should do better about that, though.
-Greg

On Mon, Oct 30, 2017, 11:40 AM Mario Giammarco <mgiammarco@xxxxxxxxx> wrote:

 >[Questions to the list]
 >How is it possible that the cluster cannot repair itself with ceph pg
repair?
 >No good copies are remaining?
 >Cannot decide which copy is valid or up-to date?
 >If so, why not, when there is checksum, mtime for everything?
 >In this inconsistent state which object does the cluster serve when it
doesn't know which one is the valid?


I am asking the same questions too, it seems strange to me that in a
fault tolerant clustered file storage like Ceph there is no
documentation about this.

I know that I am pedantic but please note that saying "to be sure use
three copies" is not enough because I am not sure what Ceph really does
when three copies are not matching.





_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com