Hello. I have noticed that during the startup phase (block check) one single bitcoin core instance is sufficient to put the "active" glusterfs server at 100% CPU for several minutes. This is not happening even with a number of parallel "dd if=/dev/zero of=/glustermointpoint/files", or other file intensive applications, and the performance problems are big for all other clients (even a simple ls hangs for long time). I'm aware that the bitcoin core is opening a really large number of files (+- 2000 pro session usually here) however I don't have the details on what exactly the bitcoin core is doing during his startup. My test environment is composed of two KVM hosts, connected using a dedicated vnet, serving as replication servers + 1 physical system on a 100mbit ethernet used only for backups with distributed volumes. The gluster volumes are mounted from the clients (KVM machines) with a "mount -t glusterfs". Does anyone else have noticed a such issue with the bitcoin core and/or has and ideas on what I could verify/monitor to solve/understand this problem? Tnx in advance for help. Marco _______________________________________________ Gluster-users mailing list Gluster-users@xxxxxxxxxxx http://www.gluster.org/mailman/listinfo/gluster-users
