I’d say if it is not round-robined, same hosts will go to the same endpoints and you can end up in an unbalanced situation. Istvan Szabo Senior Infrastructure Engineer --------------------------------------------------- Agoda Services Co., Ltd. e: istvan.szabo@xxxxxxxxx<mailto:istvan.szabo@xxxxxxxxx> --------------------------------------------------- On 2021. Aug 14., at 16:57, mhnx <morphinwithyou@xxxxxxxxx> wrote: Email received from the internet. If in doubt, don't click any link nor open any attachment ! ________________________________ I use hardware loadbalancer. I think it was Round-robin but I'm not sure. When I listen requests, rgw usages seems equal. What is your point by asking loadbalancer? If you think that the leaked rgw is using more its not the case you are looking for. I couldnt find any difference. Rgw's and hosts are identical. I've restarted that rgw and memory usage is gone. I will be watching all Rgw's to understand beter. I give +1 vote, its probably memory leak. 14 Ağu 2021 Cmt 09:56 tarihinde Szabo, Istvan (Agoda) <Istvan.Szabo@xxxxxxxxx<mailto:Istvan.Szabo@xxxxxxxxx>> şunu yazdı: Are you using loadbalancer? Maybe you use source based balancing method? Istvan Szabo Senior Infrastructure Engineer --------------------------------------------------- Agoda Services Co., Ltd. e: istvan.szabo@xxxxxxxxx<mailto:istvan.szabo@xxxxxxxxx> --------------------------------------------------- On 2021. Aug 13., at 16:14, mhnx <morphinwithyou@xxxxxxxxx<mailto:morphinwithyou@xxxxxxxxx>> wrote: Email received from the internet. If in doubt, don't click any link nor open any attachment ! ________________________________ Hello Martin. I'm using 14.2.16 Our S3 usage is similar. I have 10 rgw. They're running on OSD nodes. 9 RGW is between 3G and 5G. But One of my rgw is using 85G. I have 256G ram and that's why I didn't see that before. Thanks for the warning. But the question is Why 9 RGW is low and one of them very high? Weird... My ceph.conf: [radosgw.9] rgw data = /var/lib/ceph/radosgw/ceph-radosgw.9 rgw zonegroup = xx rgw zone = xxx rgw zonegroup root pool = xx.root rgw zone root pool = xx.root host = HOST9 rgw dns name = s3.xxxx.com<http://s3.xxxx.com> rgw frontends = beast port=8010 rgw user max buckets=999 log file = /var/log/ceph/radosgw.9.log rgw run sync thread = false rgw_dynamic_resharding = false Martin Traxl <martin.traxl@xxxxxxxx<mailto:martin.traxl@xxxxxxxx>>, 13 Ağu 2021 Cum, 14:53 tarihinde şunu yazdı: We are experiencing this behaviour eversince this cluster is productive and gets "some load". We started with this cluster in May this year, running Ceph 14.2.15 and already had this same issue. It just took a little longer until all RAM was consumed, as the load was a little lower than it is now. This is my config diff (I stripped some hostnames/IPs): { "diff": { "admin_socket": { "default": "$run_dir/$cluster-$name.$pid.$cctid.asok", "final": "/var/run/ceph/ceph-client.rgw.#####.882549.94336165049544.asok" }, "bluefs_buffered_io": { "default": true, "file": true, "final": true }, "cluster_network": { "default": "", "file": "#####/26", "final": "#####/26" }, "daemonize": { "default": true, "override": false, "final": false }, "debug_rgw": { "default": "1/5", "final": "1/5" }, "filestore_fd_cache_size": { "default": 128, "file": 2048, "final": 2048 }, "filestore_op_threads": { "default": 2, "file": 8, "final": 8 }, "filestore_queue_max_ops": { "default": 50, "file": 100, "final": 100 }, "fsid": { "default": "00000000-0000-0000-0000-000000000000", "file": "#####", "override": "#####", "final": "#####" }, "keyring": { "default": "$rgw_data/keyring", "final": "/var/lib/ceph/radosgw/ceph-rgw.#####/keyring" }, "mon_host": { "default": "", "file": "##### ##### #####", "final": "##### ##### #####" }, "mon_osd_down_out_interval": { "default": 600, "file": 1800, "final": 1800 }, "mon_osd_down_out_subtree_limit": { "default": "rack", "file": "host", "final": "host" }, "mon_osd_initial_require_min_compat_client": { "default": "jewel", "file": "jewel", "final": "jewel" }, "mon_osd_min_down_reporters": { "default": 2, "file": 2, "final": 2 }, "mon_osd_reporter_subtree_level": { "default": "host", "file": "host", "final": "host" }, "ms_client_mode": { "default": "crc secure", "file": "secure", "final": "secure" }, "ms_cluster_mode": { "default": "crc secure", "file": "secure", "final": "secure" }, "ms_mon_client_mode": { "default": "secure crc", "file": "secure", "final": "secure" }, "ms_mon_cluster_mode": { "default": "secure crc", "file": "secure", "final": "secure" }, "ms_mon_service_mode": { "default": "secure crc", "file": "secure", "final": "secure" }, "ms_service_mode": { "default": "crc secure", "file": "secure", "final": "secure" }, "objecter_inflight_ops": { "default": 24576, "final": 24576 }, "osd_backfill_scan_max": { "default": 512, "file": 16, "final": 16 }, "osd_backfill_scan_min": { "default": 64, "file": 8, "final": 8 }, "osd_deep_scrub_stride": { "default": "524288", "file": "1048576", "final": "1048576" }, "osd_fast_shutdown": { "default": true, "file": false, "final": false }, "osd_heartbeat_min_size": { "default": "2000", "file": "0", "final": "0" }, "osd_journal_size": { "default": "5120", "file": "4096", "final": "4096" }, "osd_max_backfills": { "default": 1, "file": 1, "final": 1 }, "osd_max_scrubs": { "default": 1, "file": 1, "final": 1 }, "osd_op_complaint_time": { "default": 30, "file": 5, "final": 5 }, "osd_pool_default_flag_hashpspool": { "default": true, "file": true, "final": true }, "osd_pool_default_min_size": { "default": 0, "file": 1, "final": 1 }, "osd_pool_default_size": { "default": 3, "file": 3, "final": 3 }, "osd_recovery_max_active": { "default": 3, "file": 1, "final": 1 }, "osd_recovery_max_single_start": { "default": 1, "file": 1, "final": 1 }, "osd_recovery_op_priority": { "default": 3, "file": 3, "final": 3 }, "osd_recovery_sleep_hdd": { "default": 0.10000000000000001, "file": 0, "final": 0 }, "osd_scrub_begin_hour": { "default": 0, "file": 5, "final": 5 }, "osd_scrub_chunk_max": { "default": 25, "file": 1, "final": 1 }, "osd_scrub_chunk_min": { "default": 5, "file": 1, "final": 1 }, "osd_recovery_op_priority": { "default": 3, "file": 3, "final": 3 }, "osd_recovery_sleep_hdd": { "default": 0.10000000000000001, "file": 0, "final": 0 }, "osd_scrub_begin_hour": { "default": 0, "file": 5, "final": 5 }, "osd_scrub_chunk_max": { "default": 25, "file": 1, "final": 1 }, "osd_scrub_chunk_min": { "default": 5, "file": 1, "final": 1 }, "osd_scrub_during_recovery": { "default": false, "file": true, "final": true }, "osd_scrub_end_hour": { "default": 24, "file": 23, "final": 23 }, "osd_scrub_load_threshold": { "default": 0.5, "file": 1, "final": 1 }, "osd_scrub_priority": { "default": 5, "file": 1, "final": 1 }, "osd_snap_trim_priority": { "default": 5, "file": 1, "final": 1 }, "osd_snap_trim_sleep": { "default": 0, "file": 1, "final": 1 }, "public_network": { "default": "", "file": "#####/26", "final": "#####/26" }, "rbd_default_features": { "default": "61", "final": "61" }, "rgw_dns_name": { "default": "", "file": "#####", "final": "#####" }, "rgw_frontends": { "default": "beast port=7480", "file": "beast ssl_endpoint=#####:443 ssl_certificate=/etc/ceph/rgw-ssl/#####.pem ssl_private_key=/etc/ceph/rgw-ssl/#####.key", "final": "beast ssl_endpoint=#####:443 ssl_certificate=/etc/ceph/rgw-ssl/#####.pem ssl_private_key=/etc/ceph/rgw-ssl/#####.key" }, "rgw_ignore_get_invalid_range": { "default": false, "file": true, "final": true }, "rgw_ldap_binddn": { "default": "uid=admin,cn=users,dc=example,dc=com", "file": "uid=#####,cn=#####,cn=mf,ou=#####", "final": "uid=#####,cn=#####,cn=mf,ou=#####" }, "rgw_ldap_dnattr": { "default": "uid", "file": "uid", "final": "uid" }, "rgw_ldap_searchdn": { "default": "cn=users,cn=accounts,dc=example,dc=com", "file": "ou=#####", "final": "ou=#####" }, "rgw_ldap_secret": { "default": "/etc/openldap/secret", "file": "/etc/ceph/ldap/bindpw", "final": "/etc/ceph/ldap/bindpw" }, "rgw_ldap_uri": { "default": "ldaps://<ldap.your.domain>", "file": "ldaps://#####:636", "final": "ldaps://#####:636" }, "rgw_remote_addr_param": { "default": "REMOTE_ADDR", "file": "http_x_forwarded_for", "final": "http_x_forwarded_for" }, "rgw_s3_auth_use_ldap": { "default": false, "file": true, "final": true }, "rgw_s3_auth_use_sts": { "default": false, "file": true, "final": true }, "rgw_sts_key": { "default": "sts", "file": "#####", "final": "#####" }, "rgw_user_max_buckets": { "default": 1000, "file": -1, "final": -1 }, "setgroup": { "default": "", "cmdline": "ceph", "final": "ceph" }, "setuser": { "default": "", "cmdline": "ceph", "final": "ceph" } } } ________________________________ Von: Konstantin Shalygin <k0ste@xxxxxxxx<mailto:k0ste@xxxxxxxx>> Gesendet: Freitag, 13. August 2021 13:21 An: Martin Traxl Cc: ceph-users@xxxxxxx<mailto:ceph-users@xxxxxxx> Betreff: Re: RGW memory consumption Hi, On 13 Aug 2021, at 14:10, Martin Traxl <martin.traxl@xxxxxxxx<mailto:martin.traxl@xxxxxxxx><mailto: martin.traxl@xxxxxxxx<mailto:martin.traxl@xxxxxxxx>>> wrote: yesterday evening one of my rgw nodes died again, radosgw was killed by the kernel oom killer. [Thu Aug 12 22:10:04 2021] Out of memory: Killed process 1376 (radosgw) total-vm:70747176kB, anon-rss:63900544kB, file-rss:0kB, shmem-rss:0kB, UID:167 pgtables:131008kB oom_score_adj:0 [Thu Aug 12 22:10:09 2021] oom_reaper: reaped process 1376 (radosgw), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB The radosgw was eating up all the 64GB system memory. A few hours before this happened, mempool dump showed a total usage of only 2.1 GB of ram, while in fact radosgw was using already 84.7% of 64GB. "total": { "items": 88757980, "bytes": 2147532284 PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 1376 ceph 20 0 58.8g 52.7g 17824 S 48.2 84.7 20158:04 radosgw It seems the radowgw loses track of some memory, like there is a memory leak. Some additional information. I am running on CentOS 8.4, kernel 4.18. As already mentioned, Ceph 14.2.22. radosgw is the only notable service running on this machine. Any suggestions on this? Are there maybe any tuning settings? How could I debug this further? Please show your "config diff" from admin socket Couple of days ago I was upgraded our RGW's to 14.2.21 to 14.2.22 and don't see increase memory consumption Thanks, k _______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx<mailto:ceph-users@xxxxxxx> To unsubscribe send an email to ceph-users-leave@xxxxxxx<mailto:ceph-users-leave@xxxxxxx> _______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx<mailto:ceph-users@xxxxxxx> To unsubscribe send an email to ceph-users-leave@xxxxxxx<mailto:ceph-users-leave@xxxxxxx> _______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx To unsubscribe send an email to ceph-users-leave@xxxxxxx
 |