| Summary: | samba crashes at start | ||
|---|---|---|---|
| Product: | Mageia | Reporter: | Helge Hielscher <hhielscher> |
| Component: | RPM Packages | Assignee: | Buchan Milne <bgmilne> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | critical | ||
| Priority: | Normal | CC: | curtis_mageia, geiger.david68210, ghibomgx, mageia, marja11, pkg-bugs, shlomif |
| Version: | Cauldron | Keywords: | Triaged |
| Target Milestone: | --- | ||
| Hardware: | All | ||
| OS: | Linux | ||
| See Also: | https://bugzilla.samba.org/show_bug.cgi?id=12283 | ||
| Whiteboard: | |||
| Source RPM: | samba-4.4.5-1.mga6.src.rpm | CVE: | |
| Status comment: | |||
| Attachments: | my smb.conf | ||
|
Description
Helge Hielscher
2016-09-14 13:06:14 CEST
Created attachment 8411 [details]
my smb.conf
also crashed with smb.conf from samba-common-4.4.5-1.mga6.x86_64.rpm Thanks for the report, I believe this just needed to be rebuilt for some supporting libraries that have been updated since it was built. Please confirm that samba-4.4.5-2.mga6 no longer crashes when it reaches your mirror, and close this as FIXED if so. Unfortunately samba-common-4.4.5-2.mga6 did not fix it. I get just the same crash log. Same crash on two different computers here as well CC:
(none) =>
curtis_mageia Assigning to the registered maintainer, but also CC'ing neoclust, who has done recent active maintenance on this package and who is the registered maintainer of samba4 (whatever that may be :-þ), and CC'ing pkg-bugs ml, because the registered maintainer seems to be unavailable Keywords:
(none) =>
Triaged I confirm here too, smb daemon is not activate: # systemctl status smb â smb.service - Samba SMB Daemon Loaded: loaded (/usr/lib/systemd/system/smb.service; enabled; vendor preset: enabled) Active: failed (Result: core-dump) since mar. 2016-09-20 19:11:16 CEST; 6s ago Process: 2554 ExecStart=/usr/sbin/smbd $SMBDOPTIONS (code=dumped, signal=ABRT) Main PID: 2554 (code=dumped, signal=ABRT) Status: "smbd: ready to serve connections..." sept. 20 19:11:16 daviddavid smbd[2560]: [2016/09/20 19:11:16.024236, 0] ../lib/util/fault.c:81(fault_report) sept. 20 19:11:16 daviddavid smbd[2560]: =============================================================== sept. 20 19:11:16 daviddavid smbd[2560]: [2016/09/20 19:11:16.024254, 0] ../source3/lib/util.c:791(smb_panic_s3) sept. 20 19:11:16 daviddavid smbd[2554]: [2016/09/20 19:11:16.027501, 0] ../lib/util/fault.c:78(fault_report) sept. 20 19:11:16 daviddavid smbd[2554]: =============================================================== sept. 20 19:11:16 daviddavid smbd[2554]: [2016/09/20 19:11:16.027576, 0] ../lib/util/fault.c:79(fault_report) sept. 20 19:11:16 daviddavid smbd[2554]: INTERNAL ERROR: Signal 11 in pid 2554 (4.4.5) sept. 20 19:11:16 daviddavid systemd[1]: smb.service: Main process exited, code=dumped, status=6/ABRT sept. 20 19:11:16 daviddavid systemd[1]: smb.service: Unit entered failed state. sept. 20 19:11:16 daviddavid systemd[1]: smb.service: Failed with result 'core-dump'. # CC:
(none) =>
geiger.david68210 OK, sound the culprit was the upgrade to libtevent 0.9.30, downgrading to lib{64}tevent0-0.9.29-1.mga6 would have samba starting again. In fact 0.9.30 was pretty recent upgrade.CC:
(none) =>
ghibomgx Hi all! (In reply to Giuseppe Ghibò from comment #8) > OK, sound the culprit was the upgrade to libtevent 0.9.30, downgrading to > lib{64}tevent0-0.9.29-1.mga6 would have samba starting again. In fact 0.9.30 > was pretty recent upgrade. I can confirm that the problem disappears with the downgrade of libtevent. CC:
(none) =>
shlomif There is an upstream bug report for this: https://bugzilla.samba.org/show_bug.cgi?id=12283 I crosslinked with ours.
Nicolas Lécureuil
2016-09-22 14:40:30 CEST
See Also:
(none) =>
https://bugzilla.samba.org/show_bug.cgi?id=12283 OK, with tevent downgraded to 0.9.29, the problem disappears but now htop is showing smbd consuming 100% of a single CPU even when the samba service is completely idle. Does it also happen to you? (In reply to Shlomi Fish from comment #11) > OK, with tevent downgraded to 0.9.29, the problem disappears but now htop is > showing smbd consuming 100% of a single CPU even when the samba service is > completely idle. Does it also happen to you? Why downgraded tevent? there is a fix in samba 4.5.0 in Core/Updates_testing who works just fine now, thanks to Giuseppe :) Sorry I forgot to say here that there is samba-4.5.0-2.mga6 in updates/testing and such version works with libtevent 0.9.30 and doesn't consume all the CPUs cycles. Such version includes patches made by the samba authors. Please check too if you can (also on plain samba usage). For me it seems to work properly testing some samba service, sharing between my PCs, etc... No regression found! i will move it to release WFM/samba-4.5.0-2.mga6.src.rpm Status:
NEW =>
RESOLVED There is still one side effect: on some configuration /usr/sbin/smbd might eat 100% CPU. This was also affecting previos versions 4.4.x. To get rid of it one need to install and start cups, in particular cups and cups-filters packages and start cups (cups.service and cups-browsed.service)-. |