On Tue, Apr 1, 2014 at 12:47 PM, Shawn Webb <sw...@sourcefire.com> wrote:
> On Tue, Apr 1, 2014 at 5:30 AM, Simon Hobson <li...@thehobsons.co.uk>wrote: > >> Because I've several machines using it, I've setup one to act as a local >> server, with the others pulling their updates from it. It's been generally >> reliable for years, but since updating to 0.98.1 I'm having repeated >> problems where the slaves just stop fetching updates. >> >> As an example, one of them as of this morning was 7 revisions out of >> date. Freshclam log says : >> > main.cvd is up to date (version: 55, sigs: 2424225, f-level: 60, >> builder: neo) >> > ERROR: Can't download daily.cvd from virusdb.back.<mydomain> >> > Giving up on virusdb.back.<mydomain>... >> > Update failed. Your network may be down or none of the mirrors listed >> in /etc/clamav/freshclam.conf is working. Check > >> http://www.clamav.net/support/mirror-problem for possible reasons. >> >> Invariably, if I delete mirrors.dat and restart Freshclam it will then >> download daily.cvd : >> > main.cvd is up to date (version: 55, sigs: 2424225, f-level: 60, >> builder: neo) >> > Downloading daily.cvd [100%] >> > daily.cvd updated (version: 18725, sigs: 863475, f-level: 63, builder: >> neo) >> > bytecode.cvd is up to date (version: 236, sigs: 43, f-level: 63, >> builder: dgoddard) >> > Database updated (3287743 signatures) from virusdb.back.<mydomain> (IP: >> 172.nn.nn.nn) >> >> Systems are running Debian Wheezy and fully up to date. >> >> Checking the logs, I can see one system at 6:50 said : >> >> > ClamAV update process started at Tue Apr 1 06:50:35 2014 >> > main.cvd is up to date (version: 55, sigs: 2424225, f-level: 60, >> builder: neo) >> > Downloading daily.cvd [100%] >> > WARNING: Mirror 172.nn.nn.nn is not synchronized. >> > Trying again in 5 secs... >> > ClamAV update process started at Tue Apr 1 06:50:42 2014 >> > main.cvd is up to date (version: 55, sigs: 2424225, f-level: 60, >> builder: neo) >> > WARNING: Can't download daily.cvd from virusdb.back.<mydomain> >> > Trying again in 5 secs... >> >> And on the Apache logs of the main server, I can see daily.cvd being >> fetched at 06:50 then nothing at all after that. It looks like Freshclam >> just flags the mirror as bad and never checks it again. >> >> >> Any ideas ? > > > I suspect the fault lies in a rather small piece of code that was supposed > to make the call to recv() a little more robust. If you have the ability > (or desire) to compile from source, can you please try the attached patch? > If the patch works, I'll integrate it into our next release. > > Thanks, > > Shawn > It looks like the patch failed to attach. I've pasted the patch below: [PATCH BEGIN] diff --git a/freshclam/nonblock.c b/freshclam/nonblock.c index 8a1255e..93363d1 100644 --- a/freshclam/nonblock.c +++ b/freshclam/nonblock.c @@ -239,13 +239,6 @@ nonblock_recv (int sock, void *buf, size_t len, int flags, int secs) if (FD_ISSET(sock, &fds)) { recvd = recv(sock, buf, len, flags); - if (recvd < 0) { - if (errno == EAGAIN) - continue; - - return -1; - } - return recvd; } [PATCH END] _______________________________________________ Help us build a comprehensive ClamAV guide: https://github.com/vrtadmin/clamav-faq http://www.clamav.net/support/ml
