| 1 |
Hi! |
| 2 |
|
| 3 |
I noticed that using rsyncd's chroot() functionality (which is |
| 4 |
recommended) breaks logging if your machine is not using UTC as |
| 5 |
system time zone. |
| 6 |
|
| 7 |
The problem is this: rsyncd logs various stuff after chrooting |
| 8 |
(for example the "wrote X read Y total size Z" line. |
| 9 |
Unfortunately, at this point /etc/localtime is inaccesible, |
| 10 |
resulting in UTC times being used. Other log lines (most notably |
| 11 |
the per -session "rsync on gentoo-portage from NAME (IP)" are |
| 12 |
written *outside* the chroot(), resulting in localized times. |
| 13 |
|
| 14 |
At first I thought this might be fixed by using syslog logging, |
| 15 |
but as the syslog call is used (i.e. not syslog-over-UDP) , the |
| 16 |
problem remains. |
| 17 |
|
| 18 |
So I thought, I'd deactivate chroot()ing in order to narrow down |
| 19 |
the problem. And bing: using use chroot=false "fixes" the |
| 20 |
timestamps. |
| 21 |
|
| 22 |
Loging with broken timestamps is damn near unusable, so what |
| 23 |
should we do about it? I've written to the rsync maintainers (via |
| 24 |
their mailing list) but got no feedback whatsoever :-( |
| 25 |
|
| 26 |
Greets, |
| 27 |
Tobias |
| 28 |
|
| 29 |
PS: Seemingly, older rsyncds don't have this problem. I'm using |
| 30 |
v2.6.3 |
| 31 |
|
| 32 |
-- |
| 33 |
gentoo-mirrors@g.o mailing list |
Archives