I've an foswiki installation which is really slow (~ 60 seconds for a uncached page). I've tried to profile the installation with NYTProf, according to http://foswiki.org/Support/NYTProfDebugging with the following command:
> sudo -u www-data NYTPROF="file=/tmp/nytprof.out:addpid=1:endatexit=1" perl -wTd:NYTProf view -topic Some.Topic -username MyUsername
The script fails with an exit code 141 when I run it with profiler. If I run it without profiler (remote d:NYTProf) it exits successful and producing output.
After the profiling I've gotten a bunch of profile files in my /tmp directory:
nytprof.out.[841-1860]
But when I try to merge these files, I've get an error for the first file:
> nytprofmerge nytprof.out.*
Profile data incomplete, inflate error -5 ((null)) at end of input file, perhaps the process didn't exit cleanly or the file has been truncated (refer to TROUBLESHOOTING in the documentation)
I can merge the files without the first file, but the results are useless and shows only 87 calls to Foswiki::Sandbox::CORE:open and that's it.
Do I have any chance got get an valid profiling result? Or is there an other tool, that I can use in this case?
I'm not sure why you can't get NYTProfiler to work, we've used it to figure out some performance issues in Foswiki 2.0.2, which have been partially addressed in Foswiki 2.0.3. There are a couple of issues going on, but one major cause is our conversion to UNICODE internally, and some Perl regex issues in perl versions before 5.20. https://rt.perl.org/Public/Bug/Display.html?id=66852
Foswiki 2.0.3 made the following performance updates:
Changed some heavily called internal functions from regular expressions, to index()
Changed EditRowPlugin to generate less html that requires processing by regular expressions in the rendering module.
Made some other improvements to reduce excessive re-reading of topics.
If 2.0.3 doesn't significantly help, Check to see if the problem pages have large tables in them. If so, you might try disabling the EditRowPlugin and use EditTablePlugin.
Other than that, you might try our official support channel #foswiki on IRC, http://irclogs.foswiki.org/
The script fails with an exit code 141 when I run it with profiler.
That suggests the process received a SIGPIPE signal. The sigexit option may help.
If I run it without profiler ... it exits successful and producing output.
You're using sudo so permissions might be an issue, but that's just a guess. You'll need to dig deeper to confirm if a SIGPIPE is being received and why.
I'm not familiar with foswiki. Perhaps someone in that community could be more helpful.
Related
I'm using PostgreSQL 10 server version locally for educational purposes, and one of my tasks is to force a PANIC Error on this version just to test how is reported in the log files.
I've edited the /etc/postgresql/10/main/postgresql.conf and modified the following lines:
log_min_messages = PANIC
log_min_error_statement = PANIC
Following PostgreSQL's lastest documentation it specifies that:
PANIC Reports an error that caused all database sessions to
abort.
I wanted to know if there was an easy way to trigger this kind of error and get it to be printed on the log files.
I searched a bit, but didn't find anything that could work easily.
Setting those two parameters to PANIC is not a good idea. Leave them at ERROR or WARNING, PANIC messages will be logged anyway.
There are many ways to provoke a panic. You could for example remove write permissions on the pg_wal directory or other important data structures. Then create some data modification activity (or call pg_switch_wal a couple of times).
I'm having this weird issue: I have a program that uses CUPTI callbackAPI to monitor the kernels in the program. It runs well when it's directly launched; but when I put it under cuda-gdb and run, it failed with the following error:
error: function cuptiSubscribe(&subscriber, CUpti_CallbackFunc)my_callback, NULL) failed with error CUPTI_ERROR_NOT_INITIALIZED
I've tried all examples in CUPTI/samples and concluded that programs that use callbackAPI and activityAPI will fail under cuda-gdb. (They are all well-behaved without cuda-gdb) But the fail reason differs:
If I have calls from activityAPI, then once run it under cuda-gdb, it'll hang for a minute then exit with error:
The CUDA driver has hit an internal error. Error code: 0x100ff00000001c Further execution or debugging is unreliable. Please ensure that your temporary directory is mounted with write and exec permissions.
If I have calls from callbackAPI like my own program, then it'll fail out much sooner with the same error:
CUPTI_ERROR_NOT_INITIALIZED
Any experience on this kinda issue? I really appreciate that!
According to NVIDIA forum posting here and also referred to here, the CUDA "tools" must be used uniquely. These tools include:
CUPTI
any profiler
cuda-memcheck
a debugger
Only one of these can be "in use" on a code at a time. It should be fairly easy for developers to use a profiler, or cuda-memcheck, or a debugger independently, but a possible takeaway for those using CUPTI, who also wish to be able to use another CUDA "tool" on the same code, would be to provide a coding method to be able to disable CUPTI use in their application, when they wish to use another tool.
I want to check Selenium's coverage of my web app, which runs on mod_perl2 on CentOS 6.5.
So I installed Devel::Cover, put use Devel::Cover; in my httpd.conf's <Perl> section, and restarted Apache. It immediately writes some coverage data from my custom ErrorLogging.pm module, but then if I hit any of the app's pages via a browser, nothing further happens.
I also tried changing this in httpd.conf:
StartServers 1
MinSpareServers 1
MaxSpareServers 1
...just to make sure it'd be collecting all data from the same process. However, after restarting Apache and trying again, the result was the same.
UPDATE: I also tried launching httpd with -D ONE_PROCESS as mentioned in this thread, but the result was more or less the same, except that I had to Ctrl+C the service when done testing, because it takes over the terminal, and at that point it segfaulted. But the coverage database in the end was virtually identical.
The docs don't mention anything different that I can see. How can I get Devel::Cover to record coverage data for code execution that happens in response to actual browser requests via mod_perl2?
I came across this error that is apparently pretty common among Linux Systems.
"Too many files Open"
In my code I tried to set the Python open file limit to unlimited and it threw an error saying that I could not exceed the system limit.
import resource
try:
resource.setrlimit(resource.RLIMIT_NOFILE, (500,-1))
except Exception as err:
print err
pass
So...I Googled around a bit and followed this tutorial.
However, I set everything to 9999999 which I thought would be as close to unlimited as I could get. Now I cannot open a session as root on that machine. I can't login as root at all and am pretty much stuck. What can I do to get this machine working again? I need to be able to login as root! I am running Centos 6 and it's as up to date as possible.
Did you try turning it off and on?
If this doesn't help you can supply init=/bin/bash as kernel boot parameter to enter a root shell. Or boot from a live cd and revert your changes.
After performing an 'strace su -', I looked for the 'No such file or directory' error. When comparing the output, I found that some of those errors are ok, however, there were other files missing on my problem system that existed on a comparison system. Ultimately, it led me to a faulty line in /etc/pam.d/system-auth-ac referencing an invalid shared object.
So, my recommendation is to go through your /etc/pam.d config files and validate the existence of the shared object libraries, or, look in /var/log/secure and it should give some clue to missing shared objects as well.
I am using Fedora12.
I have downloaded eclipse-cpp-galileo-SR1-linux-gtk.
I am using Java version "1.6.0_18"
When I start eclipse, I get the following error dialog. What might be the issue?
I used to see that message when there was any error in eclipse.ini.
Try using my eclipse.ini to see if that solve the problem.
In particular, check there is no extra space at the end of each line.
That being said, the error message:
/usr/java/jre1.6.0_18/bin/../lib/i386/client/libjvm.so:
cannot enable executable stack as shared object requires: Permission denied
is probably the real issue here.
Check if this is not related to the fact that most applications are not permitted to run with an executable stack.
See this article.
This forum suggests:
#setsebool -P allow_execstack 1
After reading SELinux Reveals Bugs in other code and SELinux Reveals Bugs in other code., I then tried
#execstack -c /usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0/jre/lib/i386/client/libjvm.so
and this seemed to work also.
Update June 2012: la89ondevg mentions in the comments:
In case command above didn't made the changes so one can select to make changes by going to System->Administration->SElinux Management after that select "Boolean" and set Boolean appropriately or simply search "allow_execstack" in filter and set it.
This will work same as above command.