Hi Karl,
I don't see anything mentioning shared storage in the docs and I assume
it's now automatic but can you check 'gluster volume get all
cluster.enable-shared-storage' ?
I would give a try with RH's documentation despite it's old it has some
steps (like the shared volume) that might be needed:
https://docs.redhat.com/en/documentation/red_hat_gluster_storage/3.5/html/administration_guide/sect-preparing_to_deploy_geo-replication#Setting_Up_the_Environment_for_Geo-replication_Session
Best Regards,
Strahil Nikolov
? ?????, 16 ?????? 2024 ?. ? 01:36:24 ?. ???????+3, Karl Kleinpaste <karl
at kleinpaste.org> ??????:
I am trying to test a trivial configuration of 2 hosts, each of which has a
simple 1-brick volume, that I wish to geo-rep from one to the other.
When I first experimented with this a couple years ago, it worked, but that
effort ended prematurely and I never finished the real setup. I am coming back
to it now for other purposes.
I'm on Fedora 39 with gluster 11.1. I'm using this guide:
https://docs.gluster.org/en/main/Administrator-Guide/Geo-Replication/
Hosts are pjs and pms, with respective volumes j and n.
I have created, started, and mounted the volumes, each on their own localhost.
I can also have each mount the other's volume, so general operation seems
ok.
I have added the geoacct user and group, set up password-less login for each
local root -> geoacct at other.
gluster-mountbroker things have been done and "gluster-mountbroker
status" is sensible.
"gluster-georep-sshkey generate" has done what I'd expect.
I'm at the point of creating geo-replication. Command on pjs is:
gluster volume geo-replication j geoacct at pms::n create ssh-port 6247
push-pem
(Yes, sshd listens on a nonstandard port.)
This fails with:
Please check gsync config file. Unable to get statefile's name
geo-replication command failed
Logs, which I have been tail(1)ing as it happens, show:
Received mount request for volume j
Config file (/var/lib/glusterd/geo-replication/j_pms_n/gsyncd.conf) missing.
Looking for template config file
(/var/lib/glusterd/geo-replication/gsyncd_template.conf) [No such file or
directory]
Using default config
template(/var/lib/glusterd/geo-replication/gsyncd_template.conf).
reading data from child failed
Unable to get configuration data for j(primary), geoacct at pms::n(secondary)
from template config
Please check gsync config file. Unable to get statefile's name
Staging of operation 'Volume Geo-replication Create' failed on
localhost : Please check gsync config file. Unable to get statefile's name
When I check on the files named, a bit of unhappiness is found:
$ ls -l /var/lib/glusterd/geo-replication/j_pms_n/gsyncd.conf
/var/lib/glusterd/geo-replication/gsyncd_template.conf
ls: cannot access
'/var/lib/glusterd/geo-replication/j_pms_n/gsyncd.conf': No such file or
directory
-rwxr-xr-x 1 root geoacct 0 Aug 11 08:13
/var/lib/glusterd/geo-replication/gsyncd_template.conf
That the first is missing seems normal; geo-rep creation should instantiate it,
yes? The problem seems to be that the template file is present but empty.
How does the template file come into existence, and how should it be populated?
(~geoacct is /var/lib/glusterd/geoacct.)
Any clues would be most welcome. It feels strange to have everything working
right up to the point of geo-rep creation.
--karl
________
Community Meeting Calendar:
Schedule -
Every 2nd and 4th Tuesday at 14:30 IST / 09:00 UTC
Bridge: https://meet.google.com/cpu-eiue-hvk
Gluster-users mailing list
Gluster-users at gluster.org
https://lists.gluster.org/mailman/listinfo/gluster-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.gluster.org/pipermail/gluster-users/attachments/20240818/8f221c3d/attachment.html>