Created attachment 14224 [details] Backtrace testparm Dear Samba Team, we’re facing a problem with CTDB/Samba using the registry support. We’re using SUSE Linux Enterprise Server 12 SP3, Samba/CTDB Version is: 4.8.2-git.30.690aa93c1892.1-SUSE-SLE_12-x86_64 CTDB is up and running, but if we run testparm the program crashes with the message „PANIC: messaging not initialized“ Please find attached a backtrace and below our smb.conf and ctdbd.conf Please don't hesitate to contact me if you need more information. Kind regards Christian ##### smb.conf ####### # See smb.conf.example for a more detailed config file or # read the smb.conf manpage. # Run 'testparm' to verify the config is correct after # you modified it. [global] log level = 10 # Only these both parameters are neccessary, if runnnig samba with CTDB clustering = yes dbwrap_tdb_mutexes:* = true ctdb:registry.tdb = yes #config backend = registry include = registry ####################### ### ctdb.conf ######### CTDB_RECOVERY_LOCK=/run/gluster/shared_storage/ctdb.lock CTDB_NODES=/etc/ctdb/nodes CTDB_PUBLIC_ADDRESSES=/etc/ctdb/public_addresses CTDB_MANAGES_SAMBA=yes # Only when using samba-vfs-glusterfs! # CTDB_SAMBA_SKIP_SHARE_CHECK=yes #######################
Hi Ralph, I wrongly assigned to the tools component. Now it's all yours. :-)
Created attachment 14264 [details] testparm_dbg_output
Hello, i can confirm this issue. I upgraded my CTDB Cluster from SAMBA 4.7.7/8 to SAMBA 4.8.3. I'm also using registry support. CTDB is up and running, but throws some panic messages by using the "testparm" command. SAMBA is running on Ubuntu 18.04. smb.conf [global] clustering = yes include =registry ctdbd.conf CTDB_RECOVERY_LOCK=/cephfuse/.lockdir/ctdb.lock CTDB_SAMBA_SKIP_SHARE_CHECK=no CTDB_NODES=/usr/local/samba/etc/ctdb/nodes2 CTDB_PUBLIC_ADDRESSES=/usr/local/samba/etc/ctdb/public_addresses That's it! Please have look at my attachment. I added some hopefull helpfull debug information!
Created attachment 14281 [details] Possible patch for master Can someone test this patch? Fixes it for me on my test devel cluster. Patch should apply to 4.8 as well.
Hello, i tested your patch in samba 4.8.3. It works! Thanks for fix! Regards Micha
Thanks for the patch. I am also looking into running Samba 4.8 in a cluster. While this bugzilla only mentions testparm, the same problem also exists with at least smbcacls, smbcquotas, smbclient, rpcclient and sharesec. We probably should fix the issue for all tools.
See the linked bug under "See also" :). Fixing this is tricky, I have a WIP branch running in CI but I keep getting it wrong: <https://gitlab.com/samba-team/devel/samba/pipelines/25426325>
I tried patching a few more tools based on your patches: https://gitlab.com/samba-team/devel/samba/commits/christofschmitt-4-8-cluster https://gitlab.com/samba-team/devel/samba/pipelines/25431943
(In reply to Christof Schmitt from comment #8) Thanks! I modified your patches so messaging is initialized after popt processing is done. This is needed to ensure get_dyn_config() picks up smb.conf given via commandline argument. Also the pdbedit backbox test needed a fix to pass the testenv smb.conf to pdbedit in the test. Currently running: https://gitlab.com/samba-team/devel/samba/pipelines/25462386
*** Bug 13515 has been marked as a duplicate of this bug. ***
List of affected tools fixed by the patch for this bug: pdbedit rpcclient sharesec smbcacls smbclient smbcquotas smbstatus smbtree smbpasswd testparm
This one looks promising: https://git.samba.org/?p=slow/samba.git;a=shortlog;h=refs/heads/bug13465 Older version: https://gitlab.com/samba-team/devel/samba/pipelines/25486221 Newer version, tough only commit messages changed: https://gitlab.com/samba-team/devel/samba/pipelines/25493283
Just wondering if this is ready to go upstream and be backported?
(In reply to Martin Schwenke from comment #13) As discussed on the ML, there's still an unresolved issue with smbtorture. https://lists.samba.org/archive/samba-technical/2018-September/130177.html
(In reply to Ralph Böhme from comment #14) Created bug 13640 to track the issue with smbtorture. I propose we proceed with backporting with what we have and closing this bug. The new bug 13640 can then be used to track the remaining issue with smbtorture.
Created attachment 14511 [details] Patch for 4.9 backported from master
Created attachment 14512 [details] Patch for 4.8 backported from master
Created attachment 14513 [details] Patch for 4.7 backported from master
Comment on attachment 14513 [details] Patch for 4.7 backported from master The problem is 4.8 upwards...
Reassigning to Karolin for inclusion in 4.9 and 4.8. 4.7 is not affected by this.
(In reply to Ralph Böhme from comment #20) Pushed to autobuild-v4-{9,8-}test.
(In reply to Karolin Seeger from comment #21) Pushed to both branches. Closing out bug report. Thanks!