|Anonymous | Login | Signup for a new account||2018-04-26 11:17 CEST|
|My View | View Issues | Change Log | Roadmap | | Repositories | Search | My Account|
|View Issue Details|
|ID||Project||Category||View Status||Date Submitted||Last Update|
|0000937||bareos-core||[All Projects] file daemon||public||2018-04-08 02:50||2018-04-20 15:07|
|Target Version||Fixed in Version|
|Summary||0000937: estimate causes error if time on filedaemon and director are out of sync|
|Description||If the bareos-dir system and the bareos-fd system are not in time sync (by more than hardcoded 3 second threshold), the file daemon will generate an error message like "JobId 0: Error: getmsg.c:212 Malformed message: DIR and FD clocks differ by -35 seconds, FD automatically compensating" where the call is|
Jmsg(jcr, type, 0, _("DIR and FD clocks differ by %lld seconds, FD automatically compensating.\n"), adj);
in the file daemon from about line 1565 in dir_cmd.cc. The problem is that the director code handles this message as an exception case but cannot parse it at about line 212 in bget_dirmsg() in the director code. In my configuration, it generates an email and a log entry.
In some ways, it is helpful to get these alerts in emails, but the message could be handled more consistently like others the director expects, and it would be nice if the client were identified. Otherwise, it is guesswork figuring out which client is out of time sync. For me, there are only a few clients, but a data center might have a greater challenge. Of course, I'd expect a data center to have this sort of business much better under control...
This is low priority and mainly a nuisance. It should not be too difficult to either 1) change the calling code (level_cmd in dir_cmd.cc) to issue an error message in a format that is parsable by the catchall case in the director, or 2) add a specific case handler for time sync problem in the director code (bget_dirmsg in getmsg.cc). It would be very helpful to include the client name or backup name in the message to aid in resolving such an issue.
|Steps To Reproduce||In bconsole, run|
echo 'estimate accurate=yes level=incremental job=backup_name' | bconsole
where client "backup_name" is a system whose time is out of sync with the director by more than 3 seconds. See resulting bareos.log for the message (assuming Daemon Messages are configured to log errors).
|Tags||No tags attached.|
|2018-04-08 02:50||bozonius||New Issue|
|Copyright © 2000 - 2018 MantisBT Team|