Re: [pgsql-patches] [PATCHES] Patch to log usage of temporary files
Hi all, Sorry for arriving so late into the discussion. I don't know if it's possible but it could be useful to have the text of the query which required the creation of the temporary files as an additional DETAIL line. At least, if it's possible to have it in this part of the code. Thoughts? -- Guillaume ---(end of broadcast)--- TIP 3: Have you checked our extensive FAQ? http://www.postgresql.org/docs/faq
Re: [pgsql-patches] [PATCHES] Patch to log usage of temporary
Guillaume Smet wrote: On 1/12/07, Bruce Momjian [EMAIL PROTECTED] wrote: Usually people don't want th query unless they ask for it. One nify trick would be to print the query as DETAIL unless they are already logging queries, but that just seems too complex. If you want the query, why not just log them all? Because they can't? On a highly loaded production server, people usually don't log all the queries. Anyway, if it's too complicated to implement it, perhaps it's not worth it. I'm just curious to see how people will use this information if they don't know why the temp file was created. We have to balance functionality and ease of use. That is the way I analyze these issue. We usually wait for a few people to request additional functionality of this type, and then figure out the cleanest way to implement it. -- Bruce Momjian [EMAIL PROTECTED] EnterpriseDBhttp://www.enterprisedb.com + If your life is a hard drive, Christ can be your backup. + ---(end of broadcast)--- TIP 4: Have you searched our list archives? http://archives.postgresql.org
Re: [pgsql-patches] [PATCHES] Patch to log usage of temporary files
On 1/12/07, Bruce Momjian [EMAIL PROTECTED] wrote: Usually people don't want th query unless they ask for it. One nify trick would be to print the query as DETAIL unless they are already logging queries, but that just seems too complex. If you want the query, why not just log them all? Because they can't? On a highly loaded production server, people usually don't log all the queries. Anyway, if it's too complicated to implement it, perhaps it's not worth it. I'm just curious to see how people will use this information if they don't know why the temp file was created. -- Guillaume ---(end of broadcast)--- TIP 3: Have you checked our extensive FAQ? http://www.postgresql.org/docs/faq
Re: [pgsql-patches] [PATCHES] Patch to log usage of temporary files
In response to Guillaume Smet [EMAIL PROTECTED]: On 1/12/07, Bruce Momjian [EMAIL PROTECTED] wrote: Usually people don't want th query unless they ask for it. One nify trick would be to print the query as DETAIL unless they are already logging queries, but that just seems too complex. If you want the query, why not just log them all? Because they can't? On a highly loaded production server, people usually don't log all the queries. Anyway, if it's too complicated to implement it, perhaps it's not worth it. I'm just curious to see how people will use this information if they don't know why the temp file was created. I can only speak for myself but: * I'm already using the patch in our lab. Since the lab is the same hardware/config/etc as production, I can use the information to fine tune configs that then get migrated to production after careful testing. Since it's a lab environment, I'm free to turn on and off all sorts of stuff that would be unwise in production. Thus the lab frequently has full query logging turned on. * Currently, our production systems have plenty of spare IO. The result is that I _do_ log queries on production servers, and will continue to do so until it becomes an issue. Additionally, we have lots of room to grow with this hardware, so I can use the data collected about temp file usage to justify additional RAM. Don't know how long I'll be able to leave query logging enabled on the production systems, but I'm taking advantage of it as long as possible. * This variable can be tweaked per-session, which means if I've got queries that I suspect are causing unwarranted temp files on a production server, I can enable it on a per-connection basis to track down the problem and work on a specific query, on production systems, without too much disruption of the rest of the work that's going on: set log_temp_files = 0; run suspect query set log_temp_files = -1; investigate logs At least, those are my current plans ... -- Bill Moran Collaborative Fusion Inc. ---(end of broadcast)--- TIP 4: Have you searched our list archives? http://archives.postgresql.org