Would anyone happen to have any idea why something like this would happen? This is on Solaris 2.7 with Zope 2.1.4. I can provide any other info you like, but I'm not sure what would be useful to know. ------------------------------------- load averages: 3.21, 3.08, 2.72 13:02:57 117 processes: 111 sleeping, 2 zombie, 4 on cpu CPU states: % idle, % user, % kernel, % iowait, % swap Memory: 4096M real, 66M free, 4080M swap in use, 3255M swap free PID USERNAME THR PRI NICE SIZE RES STATE TIME CPU COMMAND 5866 zope 8 10 0 2262M 2034M cpu2 149:09 24.70% python 1983 goldtest 4 0 0 37M 32M cpu3 24:54 24.62% tag_event 2370 goldmine 4 0 1 123M 116M cpu1 9:40 23.52% perl 2894 root 1 0 0 1584K 1184K cpu0 0:00 0.44% top 2372 oracle81 1 58 0 115M 85M sleep 1:20 0.34% oracle -------------------------------------- This happens regularly, as in, every few days I have to restart Zope. When I do python drops out of top altogther. Ideas? Please? Monty
At 1:00 pm +0200 12/4/00, Monty Taylor wrote:
Would anyone happen to have any idea why something like this would happen? This is on Solaris 2.7 with Zope 2.1.4. I can provide any other info you like, but I'm not sure what would be useful to know.
------------------------------------- load averages: 3.21, 3.08, 2.72
13:02:57 117 processes: 111 sleeping, 2 zombie, 4 on cpu CPU states: % idle, % user, % kernel, % iowait, % swap Memory: 4096M real, 66M free, 4080M swap in use, 3255M swap free
PID USERNAME THR PRI NICE SIZE RES STATE TIME CPU COMMAND 5866 zope 8 10 0 2262M 2034M cpu2 149:09 24.70% python 1983 goldtest 4 0 0 37M 32M cpu3 24:54 24.62% tag_event 2370 goldmine 4 0 1 123M 116M cpu1 9:40 23.52% perl 2894 root 1 0 0 1584K 1184K cpu0 0:00 0.44% top 2372 oracle81 1 58 0 115M 85M sleep 1:20 0.34% oracle
File uploads/downloads? I think that's what's causing my own colossal memory footprints...I believe Zope is caching the files that are uploaded/downloaded, and as we have a large number of files that are > 10 megs. They're all held on external storage however. But I'm only guessing - I didn't get any concrete answers/theories as to why this might be happening when I asked this question before. hth tone ------ Dr Tony McDonald, FMCC, Networked Learning Environments Project http://nle.ncl.ac.uk/ The Medical School, Newcastle University Tel: +44 191 222 5888 Fingerprint: 3450 876D FA41 B926 D3DD F8C3 F2D0 C3B9 8B38 18A2
Tony McDonald wrote:
At 1:00 pm +0200 12/4/00, Monty Taylor wrote:
Would anyone happen to have any idea why something like this would happen? This is on Solaris 2.7 with Zope 2.1.4. I can provide any other info you like, but I'm not sure what would be useful to know.
5866 zope 8 10 0 2262M 2034M cpu2 149:09 24.70% python
File uploads/downloads?
I think that's what's causing my own colossal memory footprints...I believe Zope is caching the files that are uploaded/downloaded, and as we have a large number of files that are > 10 megs. They're all held on external storage however.
But I'm only guessing - I didn't get any concrete answers/theories as to why this might be happening when I asked this question before.
Tony, Is there anything evident from Control_Panel/manage_debug? You'll need to be a bit clever to grok the refcounting stuff, but serious memory leaks should be pretty obvious to the eye. -Michel
Tony McDonald wrote:
At 1:00 pm +0200 12/4/00, Monty Taylor wrote:
Would anyone happen to have any idea why something like this would happen? This is on Solaris 2.7 with Zope 2.1.4. I can provide any other info you like, but I'm not sure what would be useful to know.
5866 zope 8 10 0 2262M 2034M cpu2 149:09 24.70% python
File uploads/downloads?
I think that's what's causing my own colossal memory footprints...I believe Zope is caching the files that are uploaded/downloaded, and as we have a large number of files that are > 10 megs. They're all held on external storage however.
But I'm only guessing - I didn't get any concrete answers/theories as to why this might be happening when I asked this question before.
Tony,
Is there anything evident from Control_Panel/manage_debug? You'll need to be a bit clever to grok the refcounting stuff, but serious memory leaks should be pretty obvious to the eye.
-Michel
Sorry for the later reply - been out the office and am going to be away until Thursday, but this thread is too important to let lapse for a week. I've used manage_debug a little in the past - but I'm not 100% sure what it is I'm looking at. For example, the first 4 lines in my popup on one of my sites is; ExtensionClass: 266 Shared.DC.ZRDB.Results.SQLAlias: 197 DocumentTemplate.DT_Util.Eval: 193 Products.ZGadflyDA.gadfly.kjParser.ParseRule: 161 Thing is, I don't use gadfly for anything on my site. I also have 6 connection boxes open on this page - My z2.py is set for 4 threads. My other problem is that my captive audience of testers-users (the Medical Students) have all left for the Easter vac. Any thoughts on this are gratefully accepted. tone.
I can now consistently make this happen. It takes about 10 seconds. I don't know that this is the only way, but this one certainly works. I have an instance of the DemoPortal product hanging off the root of the Zope tree called Portal. It has maybe thirty extra images and a few extra pages added. If I try to export it, it seems to hang from the browser side, and Zope tries to take over the box. I know it's not just fluff in reporting, because the original way I discovered the problem was when the analysts upstairs all started reporting that they could access any of our internal web-pages, not even in Zope yet, I might add, and you see the resultant top output below. I have no idea why this action would cause this. Someone suggested sending in output from manage_debug. Which I tried to do, except that once I've shot the load on the box to hell, I can seem to load that page, and once I do, I can't save it with any useful information, because netscape reloads a page to save it's source. Any suggestions on other ways I could report data to you? Thanks alot. Monty Monty Taylor wrote:
Would anyone happen to have any idea why something like this would happen? This is on Solaris 2.7 with Zope 2.1.4. I can provide any other info you like, but I'm not sure what would be useful to know.
------------------------------------- load averages: 3.21, 3.08, 2.72 13:02:57 117 processes: 111 sleeping, 2 zombie, 4 on cpu CPU states: % idle, % user, % kernel, % iowait, % swap Memory: 4096M real, 66M free, 4080M swap in use, 3255M swap free
PID USERNAME THR PRI NICE SIZE RES STATE TIME CPU COMMAND 5866 zope 8 10 0 2262M 2034M cpu2 149:09 24.70% python 1983 goldtest 4 0 0 37M 32M cpu3 24:54 24.62% tag_event 2370 goldmine 4 0 1 123M 116M cpu1 9:40 23.52% perl 2894 root 1 0 0 1584K 1184K cpu0 0:00 0.44% top 2372 oracle81 1 58 0 115M 85M sleep 1:20 0.34% oracle --------------------------------------
This happens regularly, as in, every few days I have to restart Zope. When I do python drops out of top altogther.
It all makes sense now. Sometimes you have to check those things that you can't possibly conceive of being true. Like content managers adding 4 100+ Meg Quicktime movies to the site. This makes the memory consumption on an export of the directory and my 589M database size make sense. I knew she had added some pictures, but MY GOD! The database sits at 9M now. If Zope shoots up in memory consumption again now, I'll really panic. But for now, all seems peachy on my end. Thanks alot to everyone who looked into it. Thanks, Monty Monty Taylor wrote:
I can now consistently make this happen. It takes about 10 seconds. I don't know that this is the only way, but this one certainly works.
I have an instance of the DemoPortal product hanging off the root of the Zope tree called Portal. It has maybe thirty extra images and a few extra pages added. If I try to export it, it seems to hang from the browser side, and Zope tries to take over the box. I know it's not just fluff in reporting, because the original way I discovered the problem was when the analysts upstairs all started reporting that they could access any of our internal web-pages, not even in Zope yet, I might add, and you see the resultant top output below.
I have no idea why this action would cause this. Someone suggested sending in output from manage_debug. Which I tried to do, except that once I've shot the load on the box to hell, I can seem to load that page, and once I do, I can't save it with any useful information, because netscape reloads a page to save it's source. Any suggestions on other ways I could report data to you?
Thanks alot. Monty
Monty Taylor wrote:
Would anyone happen to have any idea why something like this would happen? This is on Solaris 2.7 with Zope 2.1.4. I can provide any other info you like, but I'm not sure what would be useful to know.
------------------------------------- load averages: 3.21, 3.08, 2.72 13:02:57 117 processes: 111 sleeping, 2 zombie, 4 on cpu CPU states: % idle, % user, % kernel, % iowait, % swap Memory: 4096M real, 66M free, 4080M swap in use, 3255M swap free
PID USERNAME THR PRI NICE SIZE RES STATE TIME CPU COMMAND 5866 zope 8 10 0 2262M 2034M cpu2 149:09 24.70% python 1983 goldtest 4 0 0 37M 32M cpu3 24:54 24.62% tag_event 2370 goldmine 4 0 1 123M 116M cpu1 9:40 23.52% perl 2894 root 1 0 0 1584K 1184K cpu0 0:00 0.44% top 2372 oracle81 1 58 0 115M 85M sleep 1:20 0.34% oracle --------------------------------------
This happens regularly, as in, every few days I have to restart Zope. When I do python drops out of top altogther.
participants (4)
-
Michel Pelletier -
Monty Taylor -
Tony McDonald -
Tony.McDonald＠newcastle.ac.uk