Pat Haley
2020-Aug-28 16:57 UTC
[Gluster-users] Removing spurious hostname from peer configuration
Hi All, We have a distributed gluster filesystem across 2 servers.? We recently realized that one of the servers (mseas-data3) has 2 hostnames for the other server (mseas-data2).? One of these is on an external port that we rarely use.? When that port went down following a power outage, we ended up in a weird state where the gluster filesystem was being served to the rest of the cluster but (a) mseas-data3 kept indicating that mseas-data2 was disconnected in response to "gluster peer status" and (b) we kept having to restart the glusterd daemon on mseas-data3.? Since we don't use the external port much and didn't think gluster used it at all it was a while before we diagnosed the problem. Now we would like to expunge that external hostname by making the following changes _Current setting on MSEAS-DATA3_ /var/lib/glusterd/peers/c1110fd9-cb99-4ca1-b18a-536a122d67ef uuid=c1110fd9-cb99-4ca1-b18a-536a122d67ef state=3 hostname1=MSEAS-DATA2.MIT.EDU hostname2=mseas-data2 _Proposed change on MSEAS-DATA3_ /var/lib/glusterd/peers/c1110fd9-cb99-4ca1-b18a-536a122d67ef uuid=c1110fd9-cb99-4ca1-b18a-536a122d67ef state=3 #hostname1=MSEAS-DATA2.MIT.EDU hostname1=mseas-data2 (manually changing a configuration file).? Is this the correct approach?? Do we need to make this change in additional files as well? Do we need to bring down the volume and daemons first? Any advice will be greatly appreciated Thanks Pat -- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Pat Haley Email: phaley at mit.edu Center for Ocean Engineering Phone: (617) 253-6824 Dept. of Mechanical Engineering Fax: (617) 253-8125 MIT, Room 5-213 http://web.mit.edu/phaley/www/ 77 Massachusetts Avenue Cambridge, MA 02139-4301 -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.gluster.org/pipermail/gluster-users/attachments/20200828/6eb24d95/attachment.html>
Strahil Nikolov
2020-Aug-28 19:20 UTC
[Gluster-users] Removing spurious hostname from peer configuration
Hi Pat I have done something similar with downtime (on a lab), but I think you don't need any downtime. Here is my idea: - Edit all nodes and remove that file - Stop Glusterd on all nodes (Glusterd , but not the bricks' processes) - Start Glusterd on all nodes It seems that if you edit on 1 node and restart , the node will get the configuration from the other node which will have the 2 hostnames. Thus , you need to power down all glusterd processes in the TSP. Yet, I haven't done it exactly like that (I stopped also the bricks, but I could afford it) and you need to do a test on a set of VMs. Best Regards, Strahil Nikolov ? ?????, 28 ?????? 2020 ?., 19:57:18 ???????+3, Pat Haley <phaley at mit.edu> ??????: Hi All, We have a distributed gluster filesystem across 2 servers.? We recently realized that one of the servers (mseas-data3) has 2 hostnames for the other server (mseas-data2).? One of these is on an external port that we rarely use.? When that port went down following a power outage, we ended up in a weird state where the gluster filesystem was being served to the rest of the cluster but (a) mseas-data3 kept indicating that mseas-data2 was disconnected in response to "gluster peer status" and (b) we kept having to restart the glusterd daemon on mseas-data3.? Since we don't use the external port much and didn't think gluster used it at all it was a while before we diagnosed the problem. Now we would like to expunge that external hostname by making the following changes Current setting on MSEAS-DATA3 /var/lib/glusterd/peers/c1110fd9-cb99-4ca1-b18a-536a122d67ef uuid=c1110fd9-cb99-4ca1-b18a-536a122d67ef state=3 hostname1=MSEAS-DATA2.MIT.EDU hostname2=mseas-data2 Proposed change on MSEAS-DATA3 /var/lib/glusterd/peers/c1110fd9-cb99-4ca1-b18a-536a122d67ef uuid=c1110fd9-cb99-4ca1-b18a-536a122d67ef state=3 #hostname1=MSEAS-DATA2.MIT.EDU hostname1=mseas-data2 (manually changing a configuration file).? Is this the correct approach?? Do we need to make this change in additional files as well? Do we need to bring down the volume and daemons first? Any advice will be greatly appreciated Thanks Pat -- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Pat Haley??????????????????????????Email:??phaley at mit.edu Center for Ocean Engineering?????? Phone:??(617) 253-6824 Dept. of Mechanical Engineering????Fax:????(617) 253-8125 MIT, Room 5-213????????????????????http://web.mit.edu/phaley/www/ 77 Massachusetts Avenue Cambridge, MA??02139-4301 ________ Community Meeting Calendar: Schedule - Every 2nd and 4th Tuesday at 14:30 IST / 09:00 UTC Bridge: https://bluejeans.com/441850968 Gluster-users mailing list Gluster-users at gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users