Re: [X2Go-User] [X2Go-Dev] Bug#471: Bug#471: ATTENTION @Mike#1 - serious bug in x2goserver upgrade script (debian wheezy+squeeze)
Further Info:
- I'm guessing it's a race condition, with two instances of x2gocleansessions running simpultaneously and locking each other up:
e.g. sample output:
ps ax|grep clean 1995 ? S 0:36 /usr/bin/perl /usr/sbin/x2gocleansessions 16142 pts/0 S+ 0:05 /usr/bin/perl /usr/sbin/x2gocleansessions
- The kill command given below in my previous e-mail doesn't work. What worked was this:
ps ax | grep x2gocleansessions | grep perl
| awk ' { print $1 }' | xargs -n 1 kill
(the backslash is so it can be broken up in two lines) You have to run this in a second terminal session (as root) while the apt-get upgrade process is stuck in the first.
It will then continue without reporting an error.
-Stefan
Am 03.04.2014 09:52, schrieb Stefan Baur:
Gah. E_NOTENOUGHCAFFEINE. This was supposed to go to the list and bugtracker as well.
-------- Original-Nachricht -------- Betreff: Re: [X2Go-Dev] Bug#471: ATTENTION @Mike#1 - serious bug in x2goserver upgrade script (debian wheezy+squeeze) Datum: Thu, 03 Apr 2014 09:50:09 +0200 Von: Stefan Baur <newsgroups.mail2@stefanbaur.de> Antwort an: Stefan Baur <newsgroups.mail2@stefanbaur.de>, 471@bugs.x2go.org, x2go-dev@lists.berlios.de An: Mike Gabriel <mike.gabriel@das-netzwerkteam.de>
Addendum:
This is part of the "ps ax" output on one of the affected machines:
2563 pts/2 S+ 0:00 apt-get upgrade -y 4958 pts/1 S+ 0:00 /usr/bin/perl -w /usr/share/debconf/frontend /var/lib/dpkg/info/x2goserver.postinst configure 4.0.1.13-0x2go1+ 4962 pts/1 Z+ 0:00 [x2goserver.post] <defunct> 5004 pts/1 S+ 0:00 /usr/bin/perl /usr/sbin/x2gocleansessions
manually killing x2gocleansessions (" kill $(pidof x2gocleansessions)") seems to be a workaround, but still, we need a real fix in the form of an updated package.
-Stefan
Am 03.04.2014 08:47, schrieb Stefan Baur:
Package: x2goserver Severity: grave Version: 4.0.1.14
Hi Mike,
this is happening on
x2goserver (4.0.1.14-0x2go1+git20140402.842+squeeze.main.1)
as well.
Please help! All my machines are stuck in the upgrade process! This is not good!
-Stefan
-------- Original-Nachricht -------- Betreff: [X2Go-User] Unable to upgrade Datum: Wed, 2 Apr 2014 19:14:53 +0300 Von: Panayotis Katsaloulis <panayotis@panayotis.com> An: x2go-user@lists.berlios.de
Hello all!
I have installed x2go, I can connect - everything is fine. In the past I think there was a problem when I tried to launch a client (as xterm session), and which I was unable to properly close, so I forced close it.
The problem is that I am unable to upgrade through apt-get upgrade though.
Now when I try to upgrade I get this message:
Processing triggers for menu (2.1.46) ... Setting up x2goserver (4.0.1.14-0x2go1+git20140402.842+wheezy.main.1) ... Group x2gouser already exists. User x2gouser already exists. [info] Cleaning up stale X2Go sessions..
… and the screen remains like this forever.
The uname -a command produces the following result:
Linux danube 3.13-1-amd64 #1 SMP Debian 3.13.7-1 (2014-03-25) x86_64 GNU/Linux
Any help?
-- Panayotis
X2Go-User mailing list X2Go-User@lists.berlios.de https://lists.berlios.de/mailman/listinfo/x2go-user
X2Go-Dev mailing list X2Go-Dev@lists.berlios.de https://lists.berlios.de/mailman/listinfo/x2go-dev
X2Go-Dev mailing list X2Go-Dev@lists.berlios.de https://lists.berlios.de/mailman/listinfo/x2go-dev
Am 03.04.2014 10:13, schrieb Stefan Baur:
ps ax | grep x2gocleansessions | grep perl
| awk ' { print $1 }' | xargs -n 1 kill
Yep, can confirm. Hangs the whole update process. Just killed the x2gocleansessions process. Btw. the awk command above did not work on Kubuntu 13.10.
Am 03.04.2014 11:51, schrieb Buddy Butterfly:
Am 03.04.2014 10:13, schrieb Stefan Baur:
ps ax | grep x2gocleansessions | grep perl
| awk ' { print $1 }' | xargs -n 1 killYep, can confirm. Hangs the whole update process. Just killed the x2gocleansessions process. Btw. the awk command above did not work on Kubuntu 13.10.
"did not work" is not a useful report.
Please specify what you did, what you expected to happen, and what happened instead (actual output, not a description of what you think it might mean).
-Stefan
Am 03.04.2014 15:41, schrieb Stefan Baur:
Am 03.04.2014 11:51, schrieb Buddy Butterfly:
Am 03.04.2014 10:13, schrieb Stefan Baur:
ps ax | grep x2gocleansessions | grep perl
| awk ' { print $1 }' | xargs -n 1 killYep, can confirm. Hangs the whole update process. Just killed the x2gocleansessions process. Btw. the awk command above did not work on Kubuntu 13.10.
"did not work" is not a useful report.
Please specify what you did, what you expected to happen, and what happened instead (actual output, not a description of what you think it might mean).
-Stefan
Your're right. Never mind. Executed the single commands....
Hi Stefan,
Am Donnerstag, 3. April 2014 schrieb Stefan Baur:
Further Info:
- I'm guessing it's a race condition, with two instances of x2gocleansessions running simpultaneously and locking each other up:
e.g. sample output:
ps ax|grep clean 1995 ? S 0:36 /usr/bin/perl /usr/sbin/x2gocleansessions 16142 pts/0 S+ 0:05 /usr/bin/perl /usr/sbin/x2gocleansessions
I found a stale dpkg process on two machines running debian wheezy, but couldn't see two x2gocleansessions scripts being executed. Doesn't that speak against the hypotheses of a race condition?
- The kill command given below in my previous e-mail doesn't work. What worked was this:
ps ax | grep x2gocleansessions | grep perl \
| awk ' { print $1 }' | xargs -n 1 kill
that awk pipeline looks a bit complicated to me. Wouldn't
kill $(pidof -x x2gocleansessions)
suffice? I guess, the command you proposed originally doesn't find any processes, as x2gocleansessions is a script file interpreted by perl. Option 'x' seems to let pidof consider also perl processes executing the x2gocleansessions script. Perhaps this is also helpfull for Buddy? For me, it worked (despite having only one x2gocleansessions script stuck).
Cheers,
Holger
4962 pts/1 Z+ 0:00 [x2goserver.post] <defunct> 5004 pts/1 S+ 0:00 /usr/bin/perl /usr/sbin/x2gocleansessions
manually killing x2gocleansessions (" kill $(pidof x2gocleansessions)") seems to be a workaround, but still, we need a real fix in the form of
On Thu, Apr 3, 2014 at 9:13 AM, Holger Krause <Holger.Krause@med.uni-duesseldorf.de> wrote:
- The kill command given below in my previous e-mail doesn't work. What worked was this:
ps ax | grep x2gocleansessions | grep perl \
| awk ' { print $1 }' | xargs -n 1 kill
that awk pipeline looks a bit complicated to me. Wouldn't
kill $(pidof -x x2gocleansessions)
I use the rpm version and haven't had this issue (yet...) but please, whoever works on the update scripts - please do not kill the session running the updates. When using NX/freenx in a similar situation I have had the session where yum was running killed mid-update leaving the rpm and yum databases horribly confused and it took a day of fiddling to get things working again.
-- Les Mikesell lesmikesell@gmail.com
-
Buddy Butterfly -
Holger Krause -
Les Mikesell -
Stefan Baur