Re: [Asterisk-Users] asterisk logger - urgent!!!
On Sunday 12 Feb 2006 01:04, Kevin P. Fleming wrote: [...] (and by 'you' I don't mean you specifically, Michael, I am referring to those in this thread who think we should change the code used to write to every file we write to in Asterisk) I assume you are referring to me. I did not say that. Using write for cdr/cdr.c is perfectly feasible and as it does not use fprintf and instead uses a fixed buffer size and would be better served by open/write. Please do not twist things around. You said it was impossible. Quite clearly it is not. B -- http://www.mailtrap.org.uk/ ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] asterisk logger - urgent!!!
On Sunday 12 Feb 2006 10:57, Bob Goddard wrote: On Sunday 12 Feb 2006 01:04, Kevin P. Fleming wrote: [...] (and by 'you' I don't mean you specifically, Michael, I am referring to those in this thread who think we should change the code used to write to every file we write to in Asterisk) I assume you are referring to me. I did not say that. Using write for cdr/cdr.c is perfectly feasible and as it does not use fprintf and instead uses a fixed buffer size and would be better served by open/write. And that should have been cdr/cdr_csv.c cdr/cdr_custom.c. -- http://www.mailtrap.org.uk/ ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] asterisk logger - urgent!!!
I agree that systems should be well-adminstered, I also prefer programs that don't run amok even when there are lapses in administration. Since detecting which file caused the SIGFSZ is impractical, how about if we do this. a) don't rotate logs on SIGFSZ if it's done it recently. b) when it does rotate files on SIGFSZ,. it should rotate the csv file, too, and any other files that are written to (maybe only of they are larger than the file size limit) Kevin P. Fleming wrote: Warren Burstein wrote: How about if it would set a global variable before each disk write so the SIGFSZ handler would know which file caused it? Ha! Signals are asynchronous. This global variable would to be lock-protected, would require copying (possibly long) paths for every write, and would not necessarily be correct when the signal arrived. Sorry, this is not a solution. There is no solution, other than paying attention to your server and making sure that files don't get ridiculously large. ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] asterisk logger - urgent!!!
Warren Burstein wrote: b) when it does rotate files on SIGFSZ,. it should rotate the csv file, too, and any other files that are written to (maybe only of they are larger than the file size limit) And there's the important point: any other files that are written to. Asterisk writes to files all over the place... voicemail recordings, channel monitor recordings, etc. ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] asterisk logger - urgent!!!
On Saturday 11 Feb 2006 00:10, Kevin P. Fleming wrote: Warren Burstein wrote: How about if it would set a global variable before each disk write so the SIGFSZ handler would know which file caused it? Ha! Signals are asynchronous. This global variable would to be lock-protected, would require copying (possibly long) paths for every write, and would not necessarily be correct when the signal arrived. Sorry, this is not a solution. There is no solution, other than paying attention to your server and making sure that files don't get ridiculously large. Well, you seem to have totally ignore what I said. Using fopen/fputs to ONLY append to a file, is quite frankly, stupid. Change it to open/write and you will be able to trap via the write return code and errno. B -- http://www.mailtrap.org.uk/ ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] asterisk logger - urgent!!!
Bob Goddard wrote: Using fopen/fputs to ONLY append to a file, is quite frankly, stupid. Change it to open/write and you will be able to trap via the write return code and errno. Patches to fix bugs are most welcome. Given that these files are written using fprintf (because they are using format strings and long lists of arguments), using write will require allocating memory, building the output there and then calling write(). Seems like an awful lot of work to avoid a simple system administration failure. ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
RE: [Asterisk-Users] asterisk logger - urgent!!!
Perhaps there's a happy medium: sprintf()? I am curious to know if putting the output into a char array with sprintf() (to preserve the output formatting) and then writing it with write(). How much additional overhead would this take? Hard to know without trying it. Is anyone in a position to write some test code that would do this and report back the results? -MC -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Kevin P. Fleming Sent: Saturday, February 11, 2006 7:39 AM To: Asterisk Users Mailing List - Non-Commercial Discussion Subject: Re: [Asterisk-Users] asterisk logger - urgent!!! Bob Goddard wrote: Using fopen/fputs to ONLY append to a file, is quite frankly, stupid. Change it to open/write and you will be able to trap via the write return code and errno. Patches to fix bugs are most welcome. Given that these files are written using fprintf (because they are using format strings and long lists of arguments), using write will require allocating memory, building the output there and then calling write(). Seems like an awful lot of work to avoid a simple system administration failure. ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] asterisk logger - urgent!!!
Michael Collins wrote: I am curious to know if putting the output into a char array with sprintf() (to preserve the output formatting) and then writing it with write(). How much additional overhead would this take? Hard to know without trying it. Please... if you don't have the skills to test things, don't make suggestions like this. This is _exactly_ what I was referring to earlier. How do you know what size array is going to be large enough? Do you just allocate an enormous one on the stack for every call to ast_log, or do you malloc() one instead (which has serious performance issues)? You can't just 'guess' what will be big enough, which then leads to running through the argument list twice... There will be a non-zero cost for doing this. Can you honestly say that _any_ of this performance penalty, no matter how small, for every user of Asterisk on every system everywhere, is worth the cost when the entire problem that caused this thread can be avoided by just paying attention to your server? No log files, CDR files or anything else in the discussion here can grow to 2GB in anything less than a few days under normal circumstances, if not far, far longer than that. If you can't be bothered to run logrotate or some sort of watching process on your server, I don't think we should be force everyone else to have lower performance just so you won't be inconvenienced :-) (and by 'you' I don't mean you specifically, Michael, I am referring to those in this thread who think we should change the code used to write to every file we write to in Asterisk) ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] asterisk logger - urgent!!!
Warren Burstein wrote: How about if it would set a global variable before each disk write so the SIGFSZ handler would know which file caused it? Ha! Signals are asynchronous. This global variable would to be lock-protected, would require copying (possibly long) paths for every write, and would not necessarily be correct when the signal arrived. Sorry, this is not a solution. There is no solution, other than paying attention to your server and making sure that files don't get ridiculously large. ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
[Asterisk-Users] asterisk logger - urgent!!!
Hi, Since yesterday my Asterisk 1.2.3 is displaying the following message every few seconds Asterisk Event Logger restartedRotated Logs Per SIGXFSZ (Exceeded file size limit) This causes my log files (verbose, queue_log) to become huge with lots of logger rotate messages, but I don't know which files is exceeding size limit, since even if I delete all log files I still get this message. Any way, I have plenty of disk space and couldn't find the reason for this message. Please help me identify the issue!Dov ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] asterisk logger - urgent!!!
On Thu, Feb 09, 2006 at 10:56:18AM -0200, Dov Bigio wrote: Hi, Since yesterday my Asterisk 1.2.3 is displaying the following message every few seconds Asterisk Event Logger restarted Rotated Logs Per SIGXFSZ (Exceeded file size limit) This causes my log files (verbose, queue_log) to become huge with lots of logger rotate messages, but I don't know which files is exceeding size limit, since even if I delete all log files I still get this message. Unrelated to the origin of the problem: Do you run 'logger reload' after deleting those logs? Otherwise Asterisk still writes to the old (deleted) logs Any way, I have plenty of disk space and couldn't find the reason for this message. Please help me identify the issue! Dov ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users -- Tzafrir Cohen | [EMAIL PROTECTED] | VIM is http://tzafrir.org.il | | a Mutt's [EMAIL PROTECTED] | | best ICQ# 16849755 | | friend ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] asterisk logger - urgent!!!
Hi Tzafrir, The problem was the file Master.csv that had reached 2.0GB. I am writing a cron script to backup this file periodically and prevent this from happening. Any way, if any developers are reading this, I don't think that rotating asterisk logs is the best way to handle this problem! Maybe a more user-friendly message could be logged, infoming which file reached the 2.0GB. About your question, yes I do, for log files. Is logger rotate could also after I delete csv files? Thank you Dov - Original Message - From: Tzafrir Cohen [EMAIL PROTECTED] To: Dov Bigio [EMAIL PROTECTED]; Asterisk Users Mailing List - Non-Commercial Discussion asterisk-users@lists.digium.com Sent: Thursday, February 09, 2006 11:42 AM Subject: Re: [Asterisk-Users] asterisk logger - urgent!!! On Thu, Feb 09, 2006 at 10:56:18AM -0200, Dov Bigio wrote: Hi, Since yesterday my Asterisk 1.2.3 is displaying the following message every few seconds Asterisk Event Logger restarted Rotated Logs Per SIGXFSZ (Exceeded file size limit) This causes my log files (verbose, queue_log) to become huge with lots of logger rotate messages, but I don't know which files is exceeding size limit, since even if I delete all log files I still get this message. Unrelated to the origin of the problem: Do you run 'logger reload' after deleting those logs? Otherwise Asterisk still writes to the old (deleted) logs Any way, I have plenty of disk space and couldn't find the reason for this message. Please help me identify the issue! Dov ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users -- Tzafrir Cohen | [EMAIL PROTECTED] | VIM is http://tzafrir.org.il | | a Mutt's [EMAIL PROTECTED] | | best ICQ# 16849755 | | friend ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] asterisk logger - urgent!!!
On Thu, Feb 09, 2006 at 11:56:02AM -0200, Dov Bigio wrote: Hi Tzafrir, The problem was the file Master.csv that had reached 2.0GB. I am writing a cron script to backup this file periodically and prevent this from happening. Any way, if any developers are reading this, I don't think that rotating asterisk logs is the best way to handle this problem! Maybe a more user-friendly message could be logged, infoming which file reached the 2.0GB. This should be the job of the log rotate script. Limit its size to 1 GB and rotate it weekly. see logrotate.conf(5) . Part of just about any linux distro. Normally run in the daily cron. About your question, yes I do, for log files. Is logger rotate could also after I delete csv files? Anybody? -- Tzafrir Cohen | [EMAIL PROTECTED] | VIM is http://tzafrir.org.il | | a Mutt's [EMAIL PROTECTED] | | best ICQ# 16849755 | | friend ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] asterisk logger - urgent!!!
Dov Bigio wrote: Any way, if any developers are reading this, I don't think that rotating asterisk logs is the best way to handle this problem! Maybe a more user-friendly message could be logged, infoming which file reached the 2.0GB. Unfortunately when we receive SIGFSZ from the kernel, we have no way to know which file caused it. The assumption in Asterisk is that the only files we write to that will ever reach that size are log files. If any other file does, there will be trouble, as you have seen. ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] asterisk logger - urgent!!!
On Thursday 09 Feb 2006 16:01, Kevin P. Fleming wrote: Dov Bigio wrote: Any way, if any developers are reading this, I don't think that rotating asterisk logs is the best way to handle this problem! Maybe a more user-friendly message could be logged, infoming which file reached the 2.0GB. Unfortunately when we receive SIGFSZ from the kernel, we have no way to know which file caused it. The assumption in Asterisk is that the only files we write to that will ever reach that size are log files. If any other file does, there will be trouble, as you have seen. Why use fputs which give you no indication of the type of error when the raw write does? B -- http://www.mailtrap.org.uk/ ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] asterisk logger - urgent!!!
Hi Kevin, I see... That's why you rotate asterisk logs everytime this message occurs.. it makes sense. Unfortunately, in my case, it was the CDR CSV files tha reached that size, so rotating logs was just worsening my situation, since asterisk started to generate rotated log files every few seconds because of that. Is there a way to rotate CDR CSV files via Asterisk, or should I handle this outside Asterisk? Thanks! Dov - Original Message - From: Kevin P. Fleming [EMAIL PROTECTED] To: Asterisk Users Mailing List - Non-Commercial Discussion asterisk-users@lists.digium.com Sent: Thursday, February 09, 2006 2:01 PM Subject: Re: [Asterisk-Users] asterisk logger - urgent!!! Dov Bigio wrote: Any way, if any developers are reading this, I don't think that rotating asterisk logs is the best way to handle this problem! Maybe a more user-friendly message could be logged, infoming which file reached the 2.0GB. Unfortunately when we receive SIGFSZ from the kernel, we have no way to know which file caused it. The assumption in Asterisk is that the only files we write to that will ever reach that size are log files. If any other file does, there will be trouble, as you have seen. ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] asterisk logger - urgent!!!
Dov Bigio wrote: Is there a way to rotate CDR CSV files via Asterisk, or should I handle this outside Asterisk? As another reply already mentioned, no and yes :-) ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] asterisk logger - urgent!!!
As of Asterisk 1.2, a Master.csv file is also created in the cdr-custom directory by default. I had a horrible time until I figured that out. Any reason this was done by default that two Master.csv files are created(cdr-csv/ and cdr-custom/)? MATT--- On 2/9/06, Dov Bigio [EMAIL PROTECTED] wrote: Hi Kevin, I see... That's why you rotate asterisk logs everytime this message occurs.. it makes sense. Unfortunately, in my case, it was the CDR CSV files tha reached that size, so rotating logs was just worsening my situation, since asterisk started to generate rotated log files every few seconds because of that. Is there a way to rotate CDR CSV files via Asterisk, or should I handle this outside Asterisk? Thanks! Dov - Original Message - From: Kevin P. Fleming [EMAIL PROTECTED] To: Asterisk Users Mailing List - Non-Commercial Discussion asterisk-users@lists.digium.com Sent: Thursday, February 09, 2006 2:01 PM Subject: Re: [Asterisk-Users] asterisk logger - urgent!!! Dov Bigio wrote: Any way, if any developers are reading this, I don't think that rotating asterisk logs is the best way to handle this problem! Maybe a more user-friendly message could be logged, infoming which file reached the 2.0GB. Unfortunately when we receive SIGFSZ from the kernel, we have no way to know which file caused it. The assumption in Asterisk is that the only files we write to that will ever reach that size are log files. If any other file does, there will be trouble, as you have seen. ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [Asterisk-Users] asterisk logger - urgent!!!
Kevin P. Fleming wrote: Dov Bigio wrote: Any way, if any developers are reading this, I don't think that rotating asterisk logs is the best way to handle this problem! Maybe a more user-friendly message could be logged, infoming which file reached the 2.0GB. Unfortunately when we receive SIGFSZ from the kernel, we have no way to know which file caused it. The assumption in Asterisk is that the only files we write to that will ever reach that size are log files. If any other file does, there will be trouble, as you have seen. How about if it would set a global variable before each disk write so the SIGFSZ handler would know which file caused it? ___ --Bandwidth and Colocation provided by Easynews.com -- Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
