System Administration
cancel
Showing results for 
Search instead for 
Did you mean: 

vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overflow.

 
KINGSLEY_1
Regular Advisor

vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overflow.

Hello All!

 

We have installed blc890i4 in our environment. We run Temenons T24 application that connects to Oracle batabase. We have the Temenos application sitting on FC  DRIVES and the database sitting on Nearline DRIVES.

 

When a teller logs in to put a transaction through there is a delay in the feed back. We also get this message 

 

" vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overflow"  in

 

/var/adm/syslog/syslog.log. Could it be that the difference in the drive speed has contributed to this?

 

Kindly help me out.

 

kingsley

12 REPLIES
Dennis Handly
Acclaimed Contributor

Re: vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overfl

>vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overflow

 

Are there any corefiles?  What is the value of maxssiz and maxssiz_64?

 

>Could it be that the difference in the drive speed has contributed to this?

 

I wouldn't think so, how would the program know that?

 

KINGSLEY_1
Regular Advisor

Re: vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overfl

max_acct_file_size                Dynamic     2560000      2560000      2560000      -         pm_acct


max_async_ports                   Dynamic     4096         4096         4096         -         asyncdsk


max_mem_window                    Dynamic     0            0            0            -         vm

 

max_thread_proc                   Dynamic     4096         4096         256          2.5%      pm_proc

 

maxdsiz                           Dynamic     4294963200   4294963200   1073741824   0.6%      vm


maxdsiz_64bit                     Dynamic     4294967296   4294967296   4294967296   12.6%     vm


maxfiles                          Static      32768        32768        2048         -         fs


maxfiles_lim                      Dynamic     63488        63488        4096         0.6%      fs


maxrsessiz                        Static      8388608      8388608      8388608      -         vm


maxrsessiz_64bit                  Static      8388608      8388608      8388608      -         vm


maxssiz                           Dynamic     401604608    401604608    8388608      0.1%      vm


maxssiz_64bit                     Dynamic     1073741824   1073741824   268435456    0.1%      vm


maxtsiz                           Dynamic     1073741824   1073741824   100663296    1.0%      vm


maxtsiz_64bit                     Dynamic     1073741824   1073741824   1073741824   0.2%      vm


maxuprc                           Dynamic     3686         3686         256          3.9%      pm_proc

Dennis Handly
Acclaimed Contributor

Re: vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overfl

maxssiz                    401604608     401604608        8388608    0.1%

maxssiz_64bit      1073741824   1073741824   268435456    0.1%

 

What command gave that info?  kcusage(1m)?  What are the column titles?

401 MB seems way too large.

KINGSLEY_1
Regular Advisor

Re: vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overfl

Hi!

 

 

Below is the kcusage output

 

Thanks

 

 #kcusage


Tunable                 Usage / Setting
=============================================
filecache_max      1126498304 / 2855484620
maxdsiz              37683200 / 4294963200
maxdsiz_64bit        46137344 / 4294967296
maxfiles_lim              304 / 63488
maxssiz                786432 / 401604608
maxssiz_64bit          720896 / 1073741824
maxtsiz              11010048 / 1073741824
maxtsiz_64bit         2162688 / 1073741824
maxuprc                   118 / 3686
max_thread_proc           101 / 4096
msgmbs                      0 / 8
msgmni                      2 / 4096
msgtql                      0 / 5120
nflocks                    75 / 8192
ninode                   2807 / 165888
nkthread                 1086 / 8416
nproc                     314 / 5000
npty                        0 / 300
nstrpty                     0 / 300
nstrtel                     1 / 60
nswapdev                    1 / 32
nswapfs                     0 / 32
semmni                     21 / 4096
semmns                    142 / 8192
shmmax               17864040 / 14032385536
shmmni                     14 / 520
shmseg                      2 / 512

KINGSLEY_1
Regular Advisor

Re: vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overfl

i have the parameters below on my DB server

 

Tunable:        filecache_max


Setting:        8160028670


                         Time                Usage           %
=============================================
Mon 10/27/14 08:30 WAT    8151859200   99.9
Mon 10/27/14 08:35 WAT    8151953408   99.9
Mon 10/27/14 08:40 WAT    8152059904   99.9
Mon 10/27/14 08:45 WAT    8152023040   99.9
Mon 10/27/14 08:50 WAT    8151957504   99.9
Mon 10/27/14 08:55 WAT    8151953408   99.9
Mon 10/27/14 09:00 WAT    8151916544   99.9
Mon 10/27/14 09:05 WAT    8151969792   99.9
Mon 10/27/14 09:10 WAT    8151867392   99.9
Mon 10/27/14 09:15 WAT    8152076288   99.9
Mon 10/27/14 09:20 WAT    8151937024   99.9
Mon 10/27/14 09:25 WAT    8151883776   99.9

Dennis Handly
Acclaimed Contributor

Re: vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overfl

>maxssiz                786432 / 401,604,608
>maxssiz_64bit          720896 / 1,073,741,824

 

You are nowhere close to getting a stack overflow, unless it is a thread stack.

KINGSLEY_1
Regular Advisor

Re: vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overfl

What about its corresponding database server which shows usage of filecache_max at 99% 

Dennis Handly
Acclaimed Contributor

Re: vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overfl

>What about its corresponding database server which shows usage of filecache_max at 99% 

 

You only care about the machine that is logging that error.

There is nothing wrong with having filecache_max at 99%, unless you have plenty of memory and can increase the value.

 

KINGSLEY_1
Regular Advisor

Re: vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overfl

Hi!

It has been observed that one particular operation that is what causes the message to pop up. We run a banking software on these servers. All other operations works without any message except VAULT MANAGEMENT SYSTEM (VMS).

 

During evenings it even becomes difficult to authorise those transactions.

 

 

About filecache_max, the value was 8GB and i increased it to 13GB, but i still get the 99%. i have a memory size of 28GB, can you recommend the highest value i can give to filecache_max? 

Dennis Handly
Acclaimed Contributor

Re: vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overfl

>About filecache_max, the value was 8GB and I increased it to 13GB, but I still get the 99%.

 

It will always expand to take all you give.  :-)

 

>I have a memory size of 28GB, can you recommend the highest value I can give to filecache_max?

 

Are you using Oracle?  A database can make better use of this memory so don't increase filecache_max that much.

KINGSLEY_1
Regular Advisor

Re: vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overfl

Ok. i would leave the filecache_max alone.

 

Concerning the "Re: vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overflow" Anytime someone logs in to our TEMENOS T24 banking application to put a transaction through, the messages start to pop up in the application server. 

 

The system get flooded and at a point in the day all the services have to be restarted before users can work again.

Dennis Handly
Acclaimed Contributor

Re: vmunix: Pid 6605 was killed due to failure in writing the signal context - possible stack overfl

>Anytime someone logs in to our TEMENOS T24 banking application to put a transaction through, the messages start to pop up in the application server. 

 

If your application is threaded, perhaps you have a thread stack overflow?

http://h30499.www3.hp.com/t5/System-Administration/Pid-lt-nnn-gt-killed-due-to-trashed-stack/m-p/5106438