What steps will reproduce the problem? 1.parallel-ssh -A -h ips.txt -l username uptime 2. 3.
What is the expected output? What do you see instead? It shows success one time but after that i get Timed out, Killed by signal 9 and my CPU get full while trying that.
What version of the product are you using? On what operating system? pssh 2.1
Please provide any additional information below. Operating system Debian Squeeze
OUTPUT: parallel-ssh -A -v -h ips.txt -l username -o /tmp/foo -e /tmp/error ps Warning: do not enter your password if anyone else has superuser privileges or access to your account. Password: [1] 09:14:34 [FAILURE] 10.211.10.10 Timed out, Killed by signal 9 [2] 09:14:34 [FAILURE] 10.211.10.72 Timed out, Killed by signal 9
Comment #1
Posted on Sep 15, 2010 by Helpful WombatPlease can you tell me whats wrong. I could to once but cannt do now. Also note that with ssh key it is OK.
Comment #2
Posted on Sep 15, 2010 by Happy CamelI've been trying to think what could cause this. You said it only happens with the "-A" option, right? Are you typing the password right away? How long does it take before you see the "Timed out, Killed by signal 9" message? Is it pretty close to instant, or is there a delay?
Comment #3
Posted on Sep 15, 2010 by Happy CamelBy the way, did you kill it, or did that happen all on its own? Does anything relevant show up in the system logs (e.g., /var/log/messages)?
Comment #4
Posted on Sep 18, 2010 by Helpful Wombati did not kill it. Interestingly i got success only once. yap there is a delay. Its take a while and then saying Timed out and signal 9. Interestingly when its trying to connect i takes hole lot of prcess of my CPU. I am using debian squeeze. I also tried on debian lenny unfortunate its the same things. you can try and also send me some success log... THanks for reply...and also waiting for more comments.....
Comment #5
Posted on Sep 20, 2010 by Happy CamelHmm. Which process is using all of the CPU? Is it pssh or ssh or pssh-askpass (running the "top" program can help if you aren't already familiar with this)? It sounds like ssh is being killed, not pssh or pssh-askpass, so I'm suspecting that the ssh process is the one using a lot of CPU.
I haven't been able to reproduce this yet, and I'm still thinking about what could be causing it; let me know if there's anything else that you think might be relevant. Thanks.
Comment #6
Posted on Nov 30, 2010 by Swift CamelI'm having a similar issue. Using the -A option, even with only a single entry in the host file results in a delay of about 60 seconds and an error: " Timed out, Killed by signal 9". top shows little CPU usage but pssh-askpass does seem to be consuming the most, albeit only ~1.3%. I'm running pssh version 2.1.1 on Ubuntu 10.04
The command is: pssh -l root -A -i -h test.txt uptime
text.txt contains only 1 line/entry: 127.0.0.1
The error and result is no different if I include a username in the hosts file or not.
TIA, Terry
Comment #7
Posted on Dec 2, 2010 by Happy CamelUnfortunately, I have tried a few times to reproduce this on Fedora without any luck. I will try to find a Ubuntu machine in a few minutes and see if I have any more success there. Do you have any other suggestions that might help me reproduce it? Thanks.
Comment #8
Posted on Dec 2, 2010 by Happy CamelOkay, I found a Ubuntu 10.04 machine and added a hosts file and ran:
parallel-ssh -l amcnabb -A -i -h test.txt whoami
but this did not timeout or hang. I'm trying to think of anything that might be different in the environment.
Do you have the same problem if you ssh into your machine (without X forwarding) and run pssh there? Is there anything if interest in your .ssh/config or in your system's ssh config?
Comment #9
Posted on Dec 18, 2010 by Happy BirdI have the same issue here. pssh 2.1.1 on Debian squeeze. (I'm using public key authentication for SSH, so it's not having to wait for a password.) Slow hosts consistently cause a timeout, Killed by signal 9.
Comment #10
Posted on Jan 8, 2011 by Happy CamelFor those who are having this problem, if you try to ssh to one of the hosts manually, do you get a "The authenticity of host 'xyz' can't be established" message? I'm still trying to find information about how to reproduce the problem. Thanks.
Comment #11
Posted on Jan 8, 2011 by Happy CamelAlso, if you specify "-x '-v'" along with "-i" do you see the following error message over and over?
debug1: read_passphrase: can't open /dev/tty: No such device or address
If this is it, then I think I may have just reproduced it.
Comment #12
Posted on Jan 8, 2011 by Happy CamelBy the way, if it's the /dev/tty thing that I described in comment #11, then I think that this can only happen if pssh is being run as root. Can anyone confirm this as well? Thanks.
Comment #13
Posted on Jan 9, 2011 by Happy CamelOkay, I think I have a solution to this in commit 8b9fb2c. Give it a try if you have a chance; if it works I would like to cut a new release just for this problem.
For those who are interested in details, it looks like the problem is that pssh didn't know what to do when the -A option was set and ssh asked the question:
""" The authenticity of host 'xyz (192.168.1.1)' can't be established. RSA key fingerprint is 33:4d:a6:09:ea:45:09:39:86:a9:cb:33:93:91:93:fe. Are you sure you want to continue connecting (yes/no)? """
Thanks, everyone, for helping me track this down, and please let me know whether or not commit 8b9fb2c solves the problem for you.
Comment #14
Posted on Jan 9, 2011 by Happy CamelSorry, commit 566a881 should actually work.
Comment #15
Posted on Jan 10, 2011 by Happy CamelAfter some further testing, I'm going to mark this as fixed. However, if you run into any further problems despite the fix, please reopen the issue. Thanks for your patience with this problem.
Status: Fixed
Labels:
Type-Defect
Priority-Medium