newlisp.org server down again

[Lutz]

http://newlisp.org">http://newlisp.org is down again. This time its more serious that I can't reboot. They (ipowerweb.com) will try a hard-reboot but want to charge for everything else.



So I might have to look for a new provider and newlisp.org will be down for some time. This will also affect my email to nuevatec.com, so only contact me on the board via private message.



Lutz

[newdep]

The GUI screenshot killed the machine ;-)

[Jeff]

https://www.nearlyfreespeech.net/">https://www.nearlyfreespeech.net/ has newLisp preinstalled :)

[Lutz]

Thanks for the ISP tips. Currently this is my situation:



Before the reboot in apache/logs/error_log many of these messages:



[Thu May 31 14:55:55 2007] [emerg] (2)No such file or directory: Child cannot open lock file: /usr/local/psa/apache/logs/httpd.lock.233



after reboot I created a /usr/local/psa/apache/logs/httpd.lock



doing:



touch httpd.lock

chown apache httpd.lock



now whenever trying to restart apache I get in the error_log:



critical_create(): semget() failed: No space left on device



I deleted about 50 of apache/logs/httpd.mm.xxxx.sem where x is a number. Now when starting apache it creates a httpd.mm.xxxx.sem but I still get in error_log:



critical_create(): semget() failed: No space left on device



HELP



Lutz





ps: running Apache 1.3

[newdep]

Thanks for the ISP tips. Currently this is my situation:



Before the reboot in apache/logs/error_log many of these messages:



[Thu May 31 14:55:55 2007] [emerg] (2)No such file or directory: Child cannot open lock file: /usr/local/psa/apache/logs/httpd.lock.233



after reboot I created a /usr/local/psa/apache/logs/httpd.lock



doing:



touch httpd.lock

chown apache httpd.lock



now whenever trying to restart apache I get in the error_log:



critical_create(): semget() failed: No space left on device



I deleted about 50 of apache/logs/httpd.mm.xxxx.sem where x is a number. Now when starting apache it creates a httpd.mm.xxxx.sem but I still get in error_log:



critical_create(): semget() failed: No space left on device



HELP



Lutz





ps: running Apache 1.3

Do you have a NSF mount or cluster-disk?







Remove the lock file that is automaticly created by "apache start" it contains the PID..



Check if there is no apache already running as i.e. Root? or is there still a Lock file around?



Any changes on the user rights?



..im thinking...

[newdep]

What does ipcs give you?



...it probably ran out of semaphore memory...

(which is very odd after a reboot)



http://www.forum.psoft.net/showthread.php?t=8428">http://www.forum.psoft.net/showthread.php?t=8428

[Lutz]

I don't want to reboot, because I am afraid the machine won't and I have to call support again. For the moment I have newLISP serving a short page.



Is there any way to free semaphore space without reboot?



Lutz

[newdep]

you have to remove the semphores with "ipcrm"



first look at what ipc are oocupied with "ipcs"



make sure no apache is running (apachectl stop..or kill and clean up the lockfiles or apache.pid lock.pid ) then do ->



then you have to remove them one by one with



ipcrm -q MSGID   (depends on yuo linux distribution i think..man ipcrm..)





But It could also be that one of the apache modules got screwed...

[newdep]

if you do a "ipcs" look for the CPID thats the PID that created the Sem.

But if there are more sessions on that system then it could very well trurn out that

there is still not enough reserved..then a kernel /proc update is needed with some

extention to /sem.  



You can determine the amount of System V IPC resources available by looking at the contents of the following files:



  /proc/sys/kernel/shmmax - The maximum size of a shared memory segment.

  /proc/sys/kernel/shmmni - The maximum number of shared memory segments.

  /proc/sys/kernel/shmall - The maximum amount of shared memory

                              that can be allocated.

  /proc/sys/kernel/sem    - The maximum number and size of semaphore sets

                              that can be allocated.



For example, to view the maximum size of a shared memory segment that can be created enter:



  cat /proc/sys/kernel/shmmax



To change the maximum size of a shared memory segment to 256 MB enter:



  echo 268435456 > /proc/sys/kernel/shmmax



Start of changeTo view the maximum number of semaphores and semaphore sets which can be created enter:



cat /proc/sys/kernel/sem



This returns 4 numbers indicating:Start of change



 SEMMSL - The maximum number of semaphores in a sempahore set

 SEMMNS - The maximum number of sempahores in the system

 SEMOPM - The maximum number of operations in a single smeop call

 SEMMNI - The maximum number of sempahore sets

[Lutz]

I am on FreeBSD, ipcs gives me:



/proc> ipcs

Message Queues:

T     ID     KEY        MODE       OWNER    GROUP



Shared Memory:

T     ID     KEY        MODE       OWNER    GROUP

m 131072    5432001 --rw------- postgres postgres



Semaphores:

T     ID     KEY        MODE       OWNER    GROUP

s  65536          0 --rw-------   apache   apache

s  65537    5432001 --rw------- postgres postgres

s  65538    5432002 --rw------- postgres postgres

s  65539    5432003 --rw------- postgres postgres

s  65540          0 --rw-------   apache   apache

s  65541          0 --rw-------   apache   apache

s  65542          0 --rw-------   apache   apache

s  65543          0 --rw-------   apache   apache

s  65544          0 --rw-------   apache   apache

s  65545          0 --rw-------   apache   apache



But using on ipcrm on any of the numbers gives me:



/proc> ipcrm -q 65536

ipcrm: msqid(65536): : Invalid argument

/proc>



This is my top:



last pid:  3264;  load averages:  0.00,  0.01,  0.00  up 0+02:30:26    17:47:47

33 processes:  1 running, 32 sleeping



Mem: 77M Active, 32M Inact, 30M Wired, 32K Cache, 34M Buf, 100M Free

Swap: 2048M Total, 2048M Free





  PID USERNAME PRI NICE  SIZE    RES STATE    TIME   WCPU    CPU COMMAND

  221 psaadm     2   0 13904K 10612K accept   0:02  0.00%  0.00% httpsd

  225 psaadm     2   0 13900K 10592K accept   0:02  0.00%  0.00% httpsd

  222 psaadm     2   0 14052K 10748K accept   0:01  0.00%  0.00% httpsd

  323 root       2   0 18364K 17996K select   0:01  0.00%  0.00% perl5.00503

  328 root       2   0 18360K 17988K select   0:01  0.00%  0.00% perl5.00503

  179 named      2   0  2604K  2256K select   0:01  0.00%  0.00% named

  224 psaadm     2   0 14040K 10752K accept   0:01  0.00%  0.00% httpsd

  223 psaadm     2   0 14048K 10744K accept   0:01  0.00%  0.00% httpsd

   90 root       2   0  2592K  1956K select   0:01  0.00%  0.00% sshd

  213 root       2   0  9412K  5756K select   0:00  0.00%  0.00% httpsd

  413 psaadm     2   0 14056K 10712K accept   0:00  0.00%  0.00% httpsd

 2811 root       2   0  1900K  1460K accept   0:00  0.00%  0.00% newlisp

  203 mysql      2   0 11936K  3012K poll     0:00  0.00%  0.00% mysqld

  453 psaadm     2   0 13132K  9792K accept   0:00  0.00%  0.00% httpsd

 3138 lutz       2   0  5292K  2240K select   0:00  0.00%  0.00% sshd

   88 root      10   0  1024K   780K nanslp   0:00  0.00%  0.00% cron

 3264 lutz      30   0  1920K  1136K RUN      0:00  0.00%  0.00% top

   79 root       2   0   984K   712K select   0:00  0.00%  0.00% syslogd



Lutz

[newdep]

"ipcrm -s SEMID"  thats the id you see with "ipcs" under SEMPHORE..



ipcrm -s 65536   should do the job...





-q if for the message queues..sorry..

[Lutz]

YES, ipcrm -s did the job, http://newlisp.org">http://newlisp.org is up again.



Thanks for saving newLISP Norman (get some sleep ;) )



Lutz

[rickyboy]

https://www.nearlyfreespeech.net/">https://www.nearlyfreespeech.net/ has newLisp preinstalled :)

Hey Jeff,

Do you use nearlyfreespeech?  And if so, how do you like it?

--Rick

[Jeff]

It's a nice service, but only PHP runs embedded in apache.  Everything else, including newLisp, is on CGI.  They do not have FastCGI, either.



That being said, its an excellent service.  I haven't used it much yet, but it is always up, fast, and you pay for the bandwidth you use.  I've been pretty impressed because they actually do only charge you by your usage - it's not a gimmick.



For a huge site that gets a lot of traffic, I don't know.  You would have to look at your actual numbers and calculate how much it would cost you to run.  I have a feeling that if you have a significant enough traffic base, it might not be cost effective.