Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I have created an Amazon EC2 micro instance, running Amazon AMI. I logged into the server using ssh client. Once it logs in successfully, if I issue the "top" command, the top output never shows up and the command never returns. It constantly waits. I have to kill the ssh session and re-login. Of course none of the other things like java, tomcat etc., are working.

I rebooted the server, same problem persists. I finally changed the instance to "small" instance, even there I am facing the same problem.

During some times of the day, it seems to work fine without the above problems.

Anybody has any ideas on why this happens? Is this related to the CPU stealing or thrashing?

SOLVED: "To avoid potential problems with MTU settings and packet loss, also add a rule to allow "All ICMP". After you create the new rule, click Apply Rule Changes." Got the solution from this link- http://code.google.com/p/opendatakit/wiki/AggregateAWSInstall

share|improve this question
    
what linux-distro, what image-type, what ami-id etc? –  Tom H May 19 '12 at 9:29
    
So to confirm, ctrl-C doesn't cause top to quit? –  Tom H May 19 '12 at 9:35
    
Ctrl-C does not cause top to quit! –  Keshav Prasad May 19 '12 at 10:26
    
Image type is Amazon Linux.. –  Keshav Prasad May 19 '12 at 10:26
    
install strace and paste the output for the hung process –  Tom H May 19 '12 at 10:34
add comment

2 Answers

up vote 4 down vote accepted

I suspect MTU problems whenever I get a suspicious network hang like this. Try cating a large text file (something over 4k), and see if that hangs the session, too. If it does, it's almost certain that you've got a small MTU somewhere along the path that's causing you issues (especially since it's time-of-day dependent; perhaps your traffic's taking a different route at different times of the day). Google around (or ask a new question) to work out how to fix MTU problems (I'm not going to spend long writing it all out here, on the off-chance I'm wrong).

share|improve this answer
    
oh, so its hanging the network connection rather than just top? –  Tom H May 19 '12 at 9:43
1  
If my suspicion is correct, yes. –  womble May 19 '12 at 10:56
    
I tried "vi" on a small file, tried "cat"ing a text file, it resulted in the same behavior. –  Keshav Prasad May 19 '12 at 13:04
    
Most likely its the MTU problem that you've mentioned! As the process which I presumed "getting stuck" was actually terminating properly when observed from another SSH session. –  Keshav Prasad May 19 '12 at 13:12
    
@womble any pointers or links to resources trying to solve the MTU problem? –  Keshav Prasad May 19 '12 at 13:19
add comment

No, but you can easily get some debugging information about what is hanging the process.

Presumably you can login to another ssh session, (or if not make sure you have 2 session opened already)

So basically if I start a long running process like so

sleep 1000

I can find it from another terminal session like so;

 # ps -ef  | grep sleep | grep -v grep
 root     11768 11287  0 10:36 pts/19   00:00:00 sleep 1000

I can examine the syscall that is being executed by that process using the strace tool (from the strace package in yum/apt repo)

# strace -f -p 11768
Process 11768 attached - interrupt to quit
restart_syscall(<... resuming interrupted call ...>^C <unfinished ...>
Process 11768 detached
share|improve this answer
    
I did "cat"ing a file on a ssh session. the "cat" did not return back with any output and got stuck. I logged in to the box again on a different ssh-terminal. When I issue "ps -ef | grep cat | grep -v grep" command, it returned nothing. This means the "cat" process actually terminated normally and hence not showing up in the process list. –  Keshav Prasad May 19 '12 at 13:10
    
its looks like @womble didn't get all those points for not knowing about stuff ... ;-) –  Tom H May 19 '12 at 13:12
    
Yep, you are right! :-) –  Keshav Prasad May 19 '12 at 13:18
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.