New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
rgw: Do not archive metadata by default #11051
Conversation
@cbodley: can you please take a look |
@prallabh this seems reasonable, Casey had described using a pool name rather than a boolean (Friday standup). |
@mattbenjamin @cbodley let me know if you suggest other means, felt this to be easier. |
@prallabh thanks for taking this on! i tend to prefer using the zone configuration to enable/disable this, for a few reasons:
my suggestion was to allow the we'd also want this to default to empty, which you've done with unfortunately, the radosgw-admin commands to change zone pool names are not very user-friendly, so we'll eventually want to look at making that easier. you currently have to manually edit the json format like so:
|
@cbodley thanks for the reasoning, can you please check if it is inline now. |
@@ -1492,7 +1492,7 @@ int RGWZoneParams::fix_pool_names() | |||
} | |||
|
|||
domain_root = fix_zone_pool_name(pool_names, name, ".rgw.data.root", domain_root.name); | |||
metadata_heap = fix_zone_pool_name(pool_names, name, ".rgw.meta", metadata_heap.name); | |||
metadata_heap = ""; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
in testing, i see that this does work to turn off the metadata_heap by default. however, it also prevents us from using zone set --infile zone.json
to turn it back on - we decode the new value for metadata_heap
, then call fix_pool_names()
and clear it again. i was able to get the zone set
to work with the following change:
- metadata_heap = fix_zone_pool_name(pool_names, name, ".rgw.meta", metadata_heap.name);
+ if (!metadata_heap.name.empty()) { // optional
+ metadata_heap = fix_zone_pool_name(pool_names, name, ".rgw.meta", metadata_heap.name);
+ }
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
sorry, my bad, I couldn't get around testing that with vstart, that said, with the above suggested change, I guess we don't need to default to empty string, isn't it?
Fixes: http://tracker.ceph.com/issues/17256 Signed-off-by: Pavan Rallabhandi <PRallabhandi@walmartlabs.com>
@cbodley can you please check now |
@cbodley was curious to know the status of this, was hoping if this could make it to 10.2.3 in time or is it too late. |
@prallabh thanks again! i've flagged http://tracker.ceph.com/issues/17256 for backport to jewel, but i'm not sure about the timing for 10.2.3 |
What does this mean for an end-user, i.e, ceph-administrator? Can I remove the meta pool and expect things to not break? |
Yeah, with this fix in one can remove the meta pool. If you are on builds that do not have this fix, one can remove the contents of the meta pool but not the pool. |
@hansbogert You can remove the pool, but if your zone's If you do want to disable the heap permanently, you can edit the zone entry to set
This needs to be documented, thanks for pointing it out - I've opened a doc bug for this at http://tracker.ceph.com/issues/18174. |
@cbodley Maybe I should address this broader: Your [@cbodley] main gripe in the mailinglist [1] was that metadata stays there even if a bucket is deleted. This resolution of optionally disabling the meta pool, IMO, does not address that issue, although having the option is of course better than the previous situation. [1] http://lists.ceph.com/pipermail/ceph-users-ceph.com/2016-September/012914.html |
@hansbogert right. after disabling the metadata_heap pool in the zone, the pool itself should be deleted to reclaim the storage |
Fixes: http://tracker.ceph.com/issues/17256
Signed-off-by: Pavan Rallabhandi PRallabhandi@walmartlabs.com