Hello, > Recently, I just noticed that there are a lot of logs about Broken pipe > error from all RGW nodes. > We set up the cluster by using Ceph-ansible script. The current version of > cluster is Octopus (15.2.13). After checking the configuration in RGW > nodes, I see that there is a config in HAProxy for sending a request to RGW > instances every 2s for health check. > > The problem is gone after disabling the check but I think this is not a > good way to fix the problem… What is the health check installed by ceph-ansible ? I’m not using ceph-ansible, but I got the same problem when it was (something like) “option httpchk GET /“. Try with “option httpchk GET /swift/healthcheck”, it solved my problem. Regards, -- Guillaume > Log: > > 2021-08-04T06:25:05.997+0000 7f4f15f7b700 1 ====== starting new request > req=0x7f4fac3d7670 ===== > > 2021-08-04T06:25:05.997+0000 7f4f15f7b700 0 ERROR: > client_io->complete_request() returned Broken pipe > > 2021-08-04T06:25:05.997+0000 7f4f15f7b700 1 ====== req done > req=0x7f4fac3d7670 op status=0 http_status=200 latency=0s ====== > > 2021-08-04T06:25:05.997+0000 7f4f15f7b700 1 beast: 0x7f4fac3d7670: > 10.0.244.246 - - [2021-08-04T06:25:05.997988+0000] "HEAD / HTTP/1.0" 200 0 > - - - > > 2021-08-04T06:25:06.337+0000 7f4ebe6cc700 1 ====== starting new request > req=0x7f4fac3d7670 ===== > > 2021-08-04T06:25:06.337+0000 7f4f13776700 0 ERROR: > client_io->complete_request() returned Broken pipe > > 2021-08-04T06:25:06.337+0000 7f4f13776700 1 ====== req done > req=0x7f4fac3d7670 op status=0 http_status=200 latency=0s ====== > > 2021-08-04T06:25:06.337+0000 7f4f13776700 1 beast: 0x7f4fac3d7670: > 10.0.244.244 - - [2021-08-04T06:25:06.337994+0000] "HEAD / HTTP/1.0" 200 0 > - - - > > 2021-08-04T06:25:07.994+0000 7f4eaa6a4700 1 ====== starting new request > req=0x7f4fac3d7670 ===== > > 2021-08-04T06:25:07.994+0000 7f4eaa6a4700 1 ====== req done > req=0x7f4fac3d7670 op status=0 http_status=200 latency=0s ====== > > 2021-08-04T06:25:07.994+0000 7f4eaa6a4700 1 beast: 0x7f4fac3d7670: > 10.0.244.245 - - [2021-08-04T06:25:07.994022+0000] "HEAD / HTTP/1.0" 200 5 > - - - > > 2021-08-04T06:25:08.002+0000 7f4ee1f13700 1 ====== starting new request > req=0x7f4fac3d7670 ===== > > 2021-08-04T06:25:08.002+0000 7f4ee1f13700 0 ERROR: > client_io->complete_request() returned Broken pipe > > 2021-08-04T06:25:08.002+0000 7f4ee1f13700 1 ====== req done > req=0x7f4fac3d7670 op status=0 http_status=200 latency=0s ====== > > 2021-08-04T06:25:08.002+0000 7f4ee1f13700 1 beast: 0x7f4fac3d7670: > 10.0.244.246 - - [2021-08-04T06:25:08.002023+0000] "HEAD / HTTP/1.0" 200 0 > - - - > > …. > > > > We set up the cluster by using Ceph-ansible script. The current version of > cluster is Octopus (15.2.13). After checking the configuration in RGW > nodes, I see that there is a config in HAProxy for sending a request to RGW > instances every 2s for health check. > > The problem is gone after disabling the check but I think this is not a > good way to fix the problem… > > > > Does anyone have experience with this problem? > > > > Many thanks, > > Nghia. > > > -- > *Thanks & Best Regards.* > > > *HP: +84935905659Email: s <trankhanh2001@xxxxxxxxx>omedayiws@xxxxxxxxx > <omedayiws@xxxxxxxxx>* > _______________________________________________ > ceph-users mailing list -- ceph-users@xxxxxxx > To unsubscribe send an email to ceph-users-leave@xxxxxxx -- Guillaume de Lafond Aqua Ray _______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx To unsubscribe send an email to ceph-users-leave@xxxxxxx
 |