[ceph-users] Re: rbd map error: couldn't connect to the cluster!
Actually I didn't try other caps. The setup of RBD images and authorizations is automised with a bash script that worked in the past w/o issues. I need to understand the root cause in order to adapt the script accordingly. Am 23.02.2023 um 17:55 schrieb Eugen Block: And did you already try the other caps? Do those work? Zitat von Thomas Schneider <74cmo...@gmail.com>: Confirmed. # ceph versions { "mon": { "ceph version 14.2.22 (877fa256043e4743620f4677e72dee5e738d1226) nautilus (stable)": 3 }, "mgr": { "ceph version 14.2.22 (877fa256043e4743620f4677e72dee5e738d1226) nautilus (stable)": 3 }, "osd": { "ceph version 14.2.22 (877fa256043e4743620f4677e72dee5e738d1226) nautilus (stable)": 437 }, "mds": { "ceph version 14.2.22 (877fa256043e4743620f4677e72dee5e738d1226) nautilus (stable)": 7 }, "overall": { "ceph version 14.2.22 (877fa256043e4743620f4677e72dee5e738d1226) nautilus (stable)": 450 } } Am 23.02.2023 um 17:33 schrieb Eugen Block: And the ceph cluster has the same version? ‚ceph versions‘ shows all daemons. If the cluster is also 14.2.X the caps should work with lower-case rbd_id. Can you confirm? Zitat von Thomas Schneider <74cmo...@gmail.com>: This is # ceph --version ceph version 14.2.22 (877fa256043e4743620f4677e72dee5e738d1226) nautilus (stable) Am 23.02.2023 um 16:47 schrieb Eugen Block: Which ceph version is this? In a Nautilus cluster it works for me with the lower-case rbd_id, in Pacific it doesn't. I don't have an Octopus cluster at hand. Zitat von Eugen Block : I tried to recreate this restrictive client access, one thing is that the rbd_id is in lower-case. I created a test client named "TEST": storage01:~ # rados -p pool ls | grep -vE "5473cdeb5c62c|1f553ba0f6222" | grep test rbd_id.test But after adding all necessary caps I'm still not allowed to get the image info: client:~ # rbd -p pool info test --id TEST --keyring /etc/ceph/ceph.client.TEST.keyring 2023-02-23T16:35:16.740+0100 7faebaffd700 -1 librbd::mirror::GetInfoRequest: 0x556072a66560 handle_get_mirror_image: failed to retrieve mirroring state: (1) Operation not permitted rbd: info: (1) Operation not permitted And I don't have rbd-mirror enabled in this cluster, so that's kind of strange... I'll try to find out which other caps it requires. I already disabled all image features but to no avail. Zitat von Thomas Schneider <74cmo...@gmail.com>: I'll delete existing authentication and its caps "VCT" and recreate it. Just to be sure: there's no ingress communication to the client (from Ceph server)? Am 23.02.2023 um 16:01 schrieb Eugen Block: For rbd commands you don't specify the "client" prefix for the --id parameter, just the client name, in your case "VCT". Your second approach shows a different error message, so it can connect with "VCT" successfully, but the permissions seem not to be sufficient. Those caps look very restrictive, not sure which prevent the map command though. Zitat von Thomas Schneider <74cmo...@gmail.com>: Hm... I'm not sure about the correct rbd command syntax, but I thought it's correct. Anyway, using a different ID fails, too: # rbd map hdb_backup/VCT --id client.VCT --keyring /etc/ceph/ceph.client.VCT.keyring rbd: couldn't connect to the cluster! # rbd map hdb_backup/VCT --id VCT --keyring /etc/ceph/ceph.client.VCT.keyring 2023-02-23T15:46:16.848+0100 7f222d19d700 -1 librbd::image::GetMetadataRequest: 0x7f220c001ef0 handle_metadata_list: failed to retrieve image metadata: (1) Operation not permitted 2023-02-23T15:46:16.848+0100 7f222d19d700 -1 librbd::image::RefreshRequest: failed to retrieve pool metadata: (1) Operation not permitted 2023-02-23T15:46:16.848+0100 7f222d19d700 -1 librbd::image::OpenRequest: failed to refresh image: (1) Operation not permitted 2023-02-23T15:46:16.848+0100 7f222c99c700 -1 librbd::ImageState: 0x5569d8a16ba0 failed to open image: (1) Operation not permitted rbd: error opening image VCT: (1) Operation not permitted Am 23.02.2023 um 15:30 schrieb Eugen Block: You don't specify which client in your rbd command: rbd map hdb_backup/VCT --id client --keyring /etc/ceph/ceph.client.VCT.keyring Have you tried this (not sure about upper-case client names, haven't tried that)? rbd map hdb_backup/VCT --id VCT --keyring /etc/ceph/ceph.client.VCT.keyring Zitat von Thomas Schneider <74cmo...@gmail.com>: Hello, I'm trying to mount RBD using rbd map, but I get this error message: # rbd map hdb_backup/VCT --id client --keyring /etc/ceph/ceph.client.VCT.keyring rbd: couldn't connect to the cluster! Checking on Ceph server the required permission for relevant keyring exists: # ceph-authtool -l /etc/ceph/ceph.client.VCT.keyring [client.VCT] key = AQBj3LZjNGn/BhAAG8IqMyH0WLKi4kTlbjiW7g== # ceph auth get client.VCT [client.VCT] key = AQBj3LZjNGn/BhAAG8IqMyH0WLKi4kTlbjiW7g== caps mon
[ceph-users] Re: rbd map error: couldn't connect to the cluster!
Just one addition from my test, I believe I misinterpreted my results because my test image was named "test" and the client "TEST", so the rbd_id. is indeed upper case for an image that has an upper-case name. So forget my comment about that. Another question though: does the image you're trying to map actually contain the object_prefix you have in your caps? Can you paste the output of 'rbd info hdb_backup/VCT'? Zitat von Thomas Schneider <74cmo...@gmail.com>: Actually I didn't try other caps. The setup of RBD images and authorizations is automised with a bash script that worked in the past w/o issues. I need to understand the root cause in order to adapt the script accordingly. Am 23.02.2023 um 17:55 schrieb Eugen Block: And did you already try the other caps? Do those work? Zitat von Thomas Schneider <74cmo...@gmail.com>: Confirmed. # ceph versions { "mon": { "ceph version 14.2.22 (877fa256043e4743620f4677e72dee5e738d1226) nautilus (stable)": 3 }, "mgr": { "ceph version 14.2.22 (877fa256043e4743620f4677e72dee5e738d1226) nautilus (stable)": 3 }, "osd": { "ceph version 14.2.22 (877fa256043e4743620f4677e72dee5e738d1226) nautilus (stable)": 437 }, "mds": { "ceph version 14.2.22 (877fa256043e4743620f4677e72dee5e738d1226) nautilus (stable)": 7 }, "overall": { "ceph version 14.2.22 (877fa256043e4743620f4677e72dee5e738d1226) nautilus (stable)": 450 } } Am 23.02.2023 um 17:33 schrieb Eugen Block: And the ceph cluster has the same version? ‚ceph versions‘ shows all daemons. If the cluster is also 14.2.X the caps should work with lower-case rbd_id. Can you confirm? Zitat von Thomas Schneider <74cmo...@gmail.com>: This is # ceph --version ceph version 14.2.22 (877fa256043e4743620f4677e72dee5e738d1226) nautilus (stable) Am 23.02.2023 um 16:47 schrieb Eugen Block: Which ceph version is this? In a Nautilus cluster it works for me with the lower-case rbd_id, in Pacific it doesn't. I don't have an Octopus cluster at hand. Zitat von Eugen Block : I tried to recreate this restrictive client access, one thing is that the rbd_id is in lower-case. I created a test client named "TEST": storage01:~ # rados -p pool ls | grep -vE "5473cdeb5c62c|1f553ba0f6222" | grep test rbd_id.test But after adding all necessary caps I'm still not allowed to get the image info: client:~ # rbd -p pool info test --id TEST --keyring /etc/ceph/ceph.client.TEST.keyring 2023-02-23T16:35:16.740+0100 7faebaffd700 -1 librbd::mirror::GetInfoRequest: 0x556072a66560 handle_get_mirror_image: failed to retrieve mirroring state: (1) Operation not permitted rbd: info: (1) Operation not permitted And I don't have rbd-mirror enabled in this cluster, so that's kind of strange... I'll try to find out which other caps it requires. I already disabled all image features but to no avail. Zitat von Thomas Schneider <74cmo...@gmail.com>: I'll delete existing authentication and its caps "VCT" and recreate it. Just to be sure: there's no ingress communication to the client (from Ceph server)? Am 23.02.2023 um 16:01 schrieb Eugen Block: For rbd commands you don't specify the "client" prefix for the --id parameter, just the client name, in your case "VCT". Your second approach shows a different error message, so it can connect with "VCT" successfully, but the permissions seem not to be sufficient. Those caps look very restrictive, not sure which prevent the map command though. Zitat von Thomas Schneider <74cmo...@gmail.com>: Hm... I'm not sure about the correct rbd command syntax, but I thought it's correct. Anyway, using a different ID fails, too: # rbd map hdb_backup/VCT --id client.VCT --keyring /etc/ceph/ceph.client.VCT.keyring rbd: couldn't connect to the cluster! # rbd map hdb_backup/VCT --id VCT --keyring /etc/ceph/ceph.client.VCT.keyring 2023-02-23T15:46:16.848+0100 7f222d19d700 -1 librbd::image::GetMetadataRequest: 0x7f220c001ef0 handle_metadata_list: failed to retrieve image metadata: (1) Operation not permitted 2023-02-23T15:46:16.848+0100 7f222d19d700 -1 librbd::image::RefreshRequest: failed to retrieve pool metadata: (1) Operation not permitted 2023-02-23T15:46:16.848+0100 7f222d19d700 -1 librbd::image::OpenRequest: failed to refresh image: (1) Operation not permitted 2023-02-23T15:46:16.848+0100 7f222c99c700 -1 librbd::ImageState: 0x5569d8a16ba0 failed to open image: (1) Operation not permitted rbd: error opening image VCT: (1) Operation not permitted Am 23.02.2023 um 15:30 schrieb Eugen Block: You don't specify which client in your rbd command: rbd map hdb_backup/VCT --id client --keyring /etc/ceph/ceph.client.VCT.keyring Have you tried this (not sure about upper-case client names, haven't tried that)? rbd map hdb_backup/VCT --id VCT --keyring /etc/ceph/ceph.client.VCT.keyring Zitat von Thomas Schneider <74cmo...@gmail.
[ceph-users] mons excessive writes to local disk and SSD wearout
Hi, on our large ceph cluster with 60 servers, 1600 OSDs, we have observed that small system nvmes are wearing out rapidly. Our monitoring shows mon writes on average about 10MB/s to store.db. For small system nvmes of 250GB and DWPD of ~1, this turns out to be too much, 0.8TB/day or 1.5PB in 5 years, too much even for 3DWPD of the same capacity. Apart from replacing the drives with larger ones, more durable, preferably both, do you have any suggestions if these writes can be reduced? Actually, the mon writes match 0.15Hz rate of .sst file creation of 64MB Best regards, Andrej -- _ prof. dr. Andrej Filipcic, E-mail: andrej.filip...@ijs.si Department of Experimental High Energy Physics - F9 Jozef Stefan Institute, Jamova 39, P.o.Box 3000 SI-1001 Ljubljana, Slovenia Tel.: +386-1-477-3674Fax: +386-1-425-7074 - ___ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-le...@ceph.io
[ceph-users] Re: rbd map error: couldn't connect to the cluster!
Please check the output here: # rbd info hdb_backup/VCT rbd image 'VCT': size 800 GiB in 204800 objects order 22 (4 MiB objects) snapshot_count: 0 id: b768d4baac048b block_name_prefix: rbd_data.b768d4baac048b format: 2 features: layering op_features: flags: create_timestamp: Thu Jan 5 15:19:14 2023 access_timestamp: Thu Jan 5 15:19:14 2023 modify_timestamp: Thu Jan 5 15:19:14 2023 Am 24.02.2023 um 11:38 schrieb Eugen Block: rbd info hdb_backup/VCT ___ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-le...@ceph.io
[ceph-users] Accessing OSD objects
Hello all, I'd really appreciate some input from the more knowledgeable here. Is there a way I can access OSD objects if I have a BlueFS replay error? This error prevents me starting the OSD and also throws an error if I try using the bluestore or objectstore tools. - I can however run a ceph-bluestore-tool show-label without issue. I'm hoping there is another way or possibly a way to purge this log that I can still access the objects on this OSD. If deleting this reply log will help (even with some data loss) I'm happy to try it. This has caused a PG to go inactive and I'm considering deleting the PG and force re-creating it. - Saw this mentioned as a last resort option. Below is a snip of where things go wrong. - I don't know if there is even a chance or is this an unrecoverable state? 2023-01-25T10:05:26.543+ 7fa773a14240 20 bluefs _replay 0x0: op_dir_link db/031664.sst to 29549 2023-01-25T10:05:26.543+ 7fa773a14240 20 bluefs _replay 0x0: op_dir_link db/031665.sst to 29550 2023-01-25T10:05:26.543+ 7fa773a14240 20 bluefs _replay 0x0: op_dir_link db/031666.sst to 29551 2023-01-25T10:05:26.543+ 7fa773a14240 20 bluefs _replay 0x0: op_dir_link db/CURRENT to 29543 2023-01-25T10:05:26.543+ 7fa773a14240 20 bluefs _replay 0x0: op_dir_link db/IDENTITY to 5 2023-01-25T10:05:26.543+ 7fa773a14240 20 bluefs _replay 0x0: op_dir_link db/LOCK to 2 2023-01-25T10:05:26.543+ 7fa773a14240 20 bluefs _replay 0x0: op_dir_link db/MANIFEST-031657 to 29542 2023-01-25T10:05:26.543+ 7fa773a14240 20 bluefs _replay 0x0: op_dir_link db/OPTIONS-031645 to 29529 2023-01-25T10:05:26.543+ 7fa773a14240 20 bluefs _replay 0x0: op_dir_link db/OPTIONS-031660 to 29545 2023-01-25T10:05:26.543+ 7fa773a14240 20 bluefs _replay 0x0: op_dir_create db.slow 2023-01-25T10:05:26.543+ 7fa773a14240 20 bluefs _replay 0x0: op_jump seq 5204712 offset 0x2 2023-01-25T10:05:26.543+ 7fa773a14240 10 bluefs _read h 0x55d2f1cfdb80 0x1~1 from file(ino 1 size 0x0 mtime 2022-10-07T17:55:34.189440+ allocated 42 alloc_commit 42 extents [1:0x177017~2,1:0x53d1e90~40]) 2023-01-25T10:05:26.543+ 7fa773a14240 20 bluefs _read left 0x1 len 0x1 2023-01-25T10:05:26.543+ 7fa773a14240 20 bluefs _read got 65536 2023-01-25T10:05:26.543+ 7fa773a14240 10 bluefs _read h 0x55d2f1cfdb80 0x2~1000 from file(ino 1 size 0x2 mtime 2022-10-07T17:55:34.189440+ allocated 42 alloc_commit 42 extents [1:0x177017~2,1:0x53d1e90~40]) 2023-01-25T10:05:26.543+ 7fa773a14240 20 bluefs _read fetching 0x0~10 of 1:0x53d1e90~40 2023-01-25T10:05:26.547+ 7fa773a14240 20 bluefs _read left 0x10 len 0x1000 2023-01-25T10:05:26.547+ 7fa773a14240 20 bluefs _read got 4096 2023-01-25T10:05:26.547+ 7fa773a14240 10 bluefs _replay 0x2: txn(seq 5204713 len 0x55 crc 0x81f48b1c) 2023-01-25T10:05:26.547+ 7fa773a14240 20 bluefs _replay 0x2: op_file_update file(ino 29551 size 0x0 mtime 2022-10-07T17:55:34.151007+ allocated 0 alloc_commit 0 extents []) 2023-01-25T10:05:26.547+ 7fa773a14240 20 bluefs _replay 0x2: op_dir_link db/031666.sst to 29551 2023-01-25T10:05:26.555+ 7fa773a14240 -1 /build/ceph-17.2.5/src/os/bluestore/BlueFS.cc: In function 'int BlueFS::_replay(bool, bool)' thread 7fa773a14240 time 2023-01-25T10:05:26.551808+ /build/ceph-17.2.5/src/os/bluestore/BlueFS.cc: 1419: FAILED ceph_assert(r == q->second->file_map.end()) Kind regards Geoff ___ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-le...@ceph.io
[ceph-users] Re: mons excessive writes to local disk and SSD wearout
Hi Andrej, That doesn't sound right -- I checked a couple of our clusters just now and the mon filesystem is writing at just a few 100kBps. debug_mon = 10 should clarify the root cause. Perhaps it's logm from some persistent slow ops? Cheers, Dan On Fri, Feb 24, 2023 at 7:36 AM Andrej Filipcic wrote: > > > Hi, > > on our large ceph cluster with 60 servers, 1600 OSDs, we have observed > that small system nvmes are wearing out rapidly. Our monitoring shows > mon writes on average about 10MB/s to store.db. For small system nvmes > of 250GB and DWPD of ~1, this turns out to be too much, 0.8TB/day or > 1.5PB in 5 years, too much even for 3DWPD of the same capacity. > > Apart from replacing the drives with larger ones, more durable, > preferably both, do you have any suggestions if these writes can be > reduced? Actually, the mon writes match 0.15Hz rate of .sst file > creation of 64MB > > Best regards, > Andrej > > -- > _ > prof. dr. Andrej Filipcic, E-mail: andrej.filip...@ijs.si > Department of Experimental High Energy Physics - F9 > Jozef Stefan Institute, Jamova 39, P.o.Box 3000 > SI-1001 Ljubljana, Slovenia > Tel.: +386-1-477-3674Fax: +386-1-425-7074 > - > ___ > ceph-users mailing list -- ceph-users@ceph.io > To unsubscribe send an email to ceph-users-le...@ceph.io ___ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-le...@ceph.io
[ceph-users] OpenSSL in librados
Hi, please forgive me if this has been asked before - I could not find any information on this topic. I am using ceph with librados via the phprados extension. Since upgrading to the current ceph versions where OpenSSL is used in in librados, I observe that using PHP's libcurl integration and other features which rely on OpenSSL randomly fail when opening a TLS connection. I suspect that librados somehow initializes or uninitializes OpenSSL in a way that interferes with the OpenSSL usage of libcurl / PHP's fsockopen. Did anybody make a similar experience? Thanks, Patrick ___ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-le...@ceph.io
[ceph-users] Large STDDEV in pg per osd
I have been digging for a while on how to minimize STDDEV of the distribution of data on my OSDs and I can't seem to get it to get below 12. I have other clusters that have a STDDEV of 1 which is my goal. But this cluster is really giving me fits. This cluster started off on the Emperor. Might have even ran Dumpling originally. I'm not sure I have some crud laying around that is preventing better balancing. I have tried the ceph balancer and I even tried jj balancer. As you can see below My lowest osd percent used is 37% and highest is 64% If anyone has any ideas on how to crack this nut I would greatly appreciate it. Thanks, Joe Here is my data: ceph osd tree ID CLASS WEIGHT TYPE NAMESTATUS REWEIGHT PRI-AFF -1 167.63680 root default -3 57.62895 rack marack -4 10.47656 host gallo 0ssd3.49219 osd.0up 1.0 1.0 1ssd3.49219 osd.1up 1.0 1.0 8ssd3.49219 osd.8up 1.0 1.0 -8 7.85712 host jeep 18ssd0.87299 osd.18 up 1.0 1.0 19ssd0.87299 osd.19 up 1.0 1.0 20ssd0.87299 osd.20 up 1.0 1.0 21ssd0.87299 osd.21 up 1.0 1.0 22ssd3.49219 osd.22 up 1.0 1.0 45ssd0.87299 osd.45 up 1.0 1.0 -5 7.85712 host joy 5ssd0.87299 osd.5up 1.0 1.0 12ssd0.87299 osd.12 up 1.0 1.0 13ssd0.87299 osd.13 up 1.0 1.0 16ssd0.87299 osd.16 up 1.0 1.0 17ssd3.49219 osd.17 up 1.0 1.0 43ssd0.87299 osd.43 up 1.0 1.0 -37 6.98625 host kc01 82ssd0.87329 osd.82 up 1.0 1.0 83ssd0.87329 osd.83 up 1.0 1.0 84ssd0.87329 osd.84 up 1.0 1.0 85ssd3.49309 osd.85 up 1.0 1.0 86ssd0.87329 osd.86 up 1.0 1.0 -50 6.98625 host ks02 97ssd0.87329 osd.97 up 1.0 1.0 98ssd0.87329 osd.98 up 1.0 1.0 99ssd0.87329 osd.99 up 1.0 1.0 100ssd0.87329 osd.100 up 1.0 1.0 101ssd3.49309 osd.101 up 1.0 1.0 -27 3.49319 host lc02 52ssd1.74660 osd.52 up 1.0 1.0 73ssd1.74660 osd.73 up 1.0 1.0 -55 13.97246 host lx01 14ssd3.49309 osd.14 up 1.0 1.0 27ssd3.49309 osd.27 up 1.0 1.0 29ssd3.49309 osd.29 up 1.0 1.0 48ssd1.74660 osd.48 up 1.0 1.0 49ssd1.74660 osd.49 up 1.0 1.0 -2 55.00392 rack marack2 -13 13.96875 host helm 11ssd3.49219 osd.11 up 1.0 1.0 28ssd3.49219 osd.28 up 1.0 1.0 30ssd3.49219 osd.30 up 1.0 1.0 31ssd3.49219 osd.31 up 1.0 1.0 -17 7.85712 host jazz 23ssd0.87299 osd.23 up 1.0 1.0 24ssd0.87299 osd.24 up 1.0 1.0 25ssd0.87299 osd.25 up 1.0 1.0 26ssd0.87299 osd.26 up 1.0 1.0 36ssd3.49219 osd.36 up 1.0 1.0 44ssd0.87299 osd.44 up 1.0 1.0 -6 6.98438 host john 2ssd3.49219 osd.2up 1.0 1.0 3ssd3.49219 osd.3up 1.0 1.0 -49 7.85712 host jolt 34ssd3.49219 osd.34 up 1.0 1.0 63ssd0.87299 osd.63 up 1.0 1.0 64ssd0.87299 osd.64 up 1.0 1.0 65ssd0.87299 osd.65 up 1.0 1.0 66ssd0.87299 osd.66 up 1.0 1.0 67ssd0.87299 osd.67 up 1.0 1.0 -19 7.85712 host juju 37ssd0.87299 osd.37 up 1.0 1.0 38ssd0.87299 osd.38 up 1.0 1.0 39ssd0.87299 osd.39 up 1.0 1.00
[ceph-users] Accessing OSD objects
Hi Anthony, thanks for reaching out. Erasure data pool (K=4, M=2) but I had more than two disk failures around the same time and the data had not fully replicated / restored elsewhere in the cluster. They are big 12TB Exos so it usually takes a few weeks to backfill / recover plus I had snaptrimming on the go. FYI - The journal's co-located on drive. Kind regards Geoff On Fri, 24 Feb 2023 at 18:30, Anthony D'Atri wrote: > Are you only doing 2 replicas? > > > > > > On Feb 24, 2023, at 08:20, Geoffrey Rhodes wrote: > > This has caused a PG to go inactive > > > ___ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-le...@ceph.io