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
Log internal unindexed searches (notes=A) #4219
Comments
Comment from mreynolds (@mreynolds389) at 2020-06-18 21:37:46 Metadata Update from @mreynolds389:
|
Comment from mreynolds (@mreynolds389) at 2020-06-18 21:38:05 Metadata Update from @mreynolds389:
|
Comment from mreynolds (@mreynolds389) at 2020-06-18 21:39:24 |
Comment from mreynolds (@mreynolds389) at 2020-06-22 15:02:58 Commit 16e47bff relates to this ticket |
Comment from mreynolds (@mreynolds389) at 2020-06-22 15:16:53
c7d81c2..9e1a88c 389-ds-base-1.4.3 -> 389-ds-base-1.4.3 |
Comment from mreynolds (@mreynolds389) at 2020-06-22 15:16:53 Metadata Update from @mreynolds389:
|
Comment from tmihinto at 2020-07-02 10:48:51 Hi Mark, I have built a patch on top of 1.3.10.1-5 using https://pagure.io/389-ds-base/c/49cbd94a0 After configuring the RHDS instance to trigger a full DB scan upon a deletion of an entry,
Running the instance under gdb gives a hint about the issue:
The function get_internal_conn_op() is used in the patch but its definition is missing. It is defined in the 1.4 branch in the file ldap/servers/slapd/util.c
The 1.3.10 branch doesn't seem to have the function definition:
Would you mind to add the missing part to the patch? Regards, |
Comment from tbordaz (@tbordaz) at 2020-07-02 11:23:05 @tmihinto this missing part comes from #2088. It allows to retrieve which client operation (op/conn) triggered the internal operation. 49029 was fixed in several commits that makes the backport a bit more complex. |
Comment from tmihinto at 2020-07-02 11:51:39 Thanks Thierry! I would prefer if Engineering can fix the patch so that when CEE folks need to build a hotfix, they don't hit the same issue ;-) Meanwhile could you please confirm that the following code changes should be enough?
Regards, |
Comment from tmihinto at 2020-07-03 14:35:54 @mreynolds389 Please let me know when the new patch is ready. Thanks, |
Comment from mreynolds (@mreynolds389) at 2020-07-03 15:38:36
Try this one:
|
Comment from tmihinto at 2020-07-03 19:35:48 @mreynolds389
|
Hey @mreynolds389, could you please push this patch to 1.3.10 branch? It is still affected by this issue. And it also made it to downstream rpms...
|
Description: In 1.3.x there is no function get_internal_conn_op() which was beign called and crashing the server. Just use the regular conn and op numbers. Fixes: #4219 Reviewed by: tmihinto(Thanks!)
Description: Add a test case. Relates: #4219 Reviewed by: @mreynolds389, @droideck, @tmihinto (Thanks!)
Cloned from Pagure issue: https://pagure.io/389-ds-base/issue/51166
Issue Description
At some point in time we stopped logging internal unindexed searches in the errors log because they were noisy and often false positives, but we have since made it more fine grained with notes=A. For a notes=A case we should log the search details in the errors log. This will help determine the root cause of some of the db lock exhaustion cases we are seeing.
The text was updated successfully, but these errors were encountered: