thanks for your attention.
temporarily with the following script I'm looking at myself.
#!/bin/bash INTERVAL=180 LOG=true function log { if $LOG; then echo $1 fi }
while true; do testx=`ps -auxw | grep "smbd" | wc -l` testy=`netstat
-ap |
grep samba | grep sock | wc -l` if test "$testx" -gt "50" -a
"$testy" -gt
"50" ;then echo "samba problem" service samba restart else
echo "samba no
problem" fi sleep $INTERVAL done
2016-09-18 2:17 GMT+03:00 Andrew Bartlett <abartlet at samba.org>:
> On Sat, 2016-09-17 at 23:26 +0300, barış tombul via samba wrote:
> > I reported the problem two months ago . I knew this would happen .
> > :((
>
> I'm sorry for the frustration. One of the hardest things about
> debugging issues like this is that it is unpredictable which clue will
> give us enough information to make progress. In this case it
> was Sander Plas who was able to get the gdb backtraces of the stuck
> processes, and those have essentially isolated the problem. (Our
> inclination is that this is a tdb mutex deadlock).
>
> Furthermore, it seems likely that this happens most often in
> production, rather than in test, as the timing of connection shutdowns
> would appear critical, so it didn't show up in 'make test'.
>
> (We also don't currently have a check for stuck processes in the test
> cleanup as we probably should)
>
> As I said, I'm sorry for your frustration, and hope we will have a fix
> to this shortly.
>
> Andrew Bartlett
> --
> Andrew Bartlett http://samba.org/~abartlet/
> Authentication Developer, Samba Team http://samba.org
> Samba Developer, Catalyst IT http://catalyst.net.nz/
> services/samba
>
>
I Had to rollback to samba 4.4.4. I have lots of services that use my samba users database. I'll wait the oficial patch an then upgrade again. Att, Enviado do meu iPhone> Em 18 de set de 2016, às 08:24, barış tombul <bbtombul at gmail.com> escreveu: > > thanks for your attention. > > temporarily with the following script I'm looking at myself. > > #!/bin/bash > INTERVAL=180 > LOG=true > function log { > if $LOG; then > echo $1 > fi > } > while true; do > testx=`ps -auxw | grep "smbd" | wc -l` > testy=`netstat -ap | grep samba | grep sock | wc -l` > if test "$testx" -gt "50" -a "$testy" -gt "50" ;then > echo "samba problem" > service samba restart > else > echo "samba no problem" > fi > sleep $INTERVAL > done > > > > 2016-09-18 2:17 GMT+03:00 Andrew Bartlett <abartlet at samba.org>: >> On Sat, 2016-09-17 at 23:26 +0300, barış tombul via samba wrote: >> > I reported the problem two months ago . I knew this would happen . >> > :(( >> >> I'm sorry for the frustration. One of the hardest things about >> debugging issues like this is that it is unpredictable which clue will >> give us enough information to make progress. In this case it >> was Sander Plas who was able to get the gdb backtraces of the stuck >> processes, and those have essentially isolated the problem. (Our >> inclination is that this is a tdb mutex deadlock). >> >> Furthermore, it seems likely that this happens most often in >> production, rather than in test, as the timing of connection shutdowns >> would appear critical, so it didn't show up in 'make test'. >> >> (We also don't currently have a check for stuck processes in the test >> cleanup as we probably should) >> >> As I said, I'm sorry for your frustration, and hope we will have a fix >> to this shortly. >> >> Andrew Bartlett >> -- >> Andrew Bartlett http://samba.org/~abartlet/ >> Authentication Developer, Samba Team http://samba.org >> Samba Developer, Catalyst IT http://catalyst.net.nz/services/samba >> >
On Sun, 2016-09-18 at 09:16 -0300, Edson Tadeu wrote:> I Had to rollback to samba 4.4.4. > > I have lots of services that use my samba users database. > > I'll wait the oficial patch an then upgrade again.This is what I would recommend. Sorry, Andrew Bartlett -- Andrew Bartlett http://samba.org/~abartlet/ Authentication Developer, Samba Team http://samba.org Samba Developer, Catalyst IT http://catalyst.net.nz/services/samba