nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

iwproxy memory consumption

System

iwserver: 5.5.2 Build 20638 Interwoven 20030604 on Windows 2000.

iwproxy has consumed almost 500MB of RAM. Could this be bugs in proxy configuration (which is minimal here), or is this a bug in the proxy itself? I remember some posts in the wayback machine about iwproxy memory leaks, but I thought they had been fixed long ago. Am I just on an obsolete build? Unfortunately it looks like it would be very hard to get any patch into production here. Is there any alternative but to just restart the service periodically?

Find more posts tagged with

Comments

plaird

As far as we know, there are no memory leaks in iwproxy in the currently supported versions of TeamSite. Often iwproxy can use a large amount of memory if it's either under heavy load or if there are recursive regexes in use. Can you give some insight into the pattern you're seeing in memory usage (what does it use when it starts? does it rise faster under load than when idle? does it suddenly spike up or is it more gradual? Also, please examine your regexes for potential looping/recursion.

Gregg Faus

john,

I think I had the same issue when my production environment was on 5.5.2 SP3. It only remedied itself when I migrated to 6.x and now it leaves just a little footprint (instead of 250+ MB).

I was never able to diagnose the real problem. I just settled for restart of the service every so often.

Migrateduser

Hey Peter, it's been a while since we worked together (Walnut Creek).

Can you tell me if the version we are on is in the list of currently supported builds?

I don't actually administer the server and the person who does doesn't seem to concerned about the consumption, they just restart the service. Since I just noticed it today, I haven't seen much pattern, but after restarting the service is now using about 2MB.

Load should be minimal.

I don't know how to get uptime on windows, but dividing summary process time by 2 CPUs it looks like it's been up for almost 30 days.

Not sure what a recursive regex is, but iw.cfg looks like this:

[iwproxy_fullproxy_redirect]
_regex=^http=http://localhost/

[iwproxy_remap]
global_default_map=/

[global_default_map]

[iwproxy_preconnect_remap]

# will send content by private message

Thanks,

-John

Adam Stoller

[iwproxy_fullproxy_redirect]
_regex=^http=http://localhost/

I don't think you want the above. First, fullproxy_redirect only works in conjunction with browsers using iwproxy.

Secondly - you'd want the source-side (left-side) of the regex to be more explicit - like:
_regex=http://mycompany.com/= http://localhost/

Check out the TeamSite Administrator's Guide (in my 5.5.2 copy it's around pages 164, 170-171, 181).

--fish
Senior Consultant, Quotient Inc.
http://www.quotient-inc.com

Migrateduser

Actually, I avoid any proxy configuration and I don't administer this server, just kind of curious about the memory leak. Nobody internally uses the Interwoven proxy as a browser proxy, but the server is on the internet and I heard there were people not associated with this organization configuring their browser proxy to use the Interwoven server to get past the Great Firewall of China. I think this line is there to reduce straing on the server from these users. I don't see how either of these lines could cause the process to use 500MB of memory - that seems like a bug. Also seems like they would be there for some reason.

Adam Stoller

I'm not sure how that proxy gets used - but as written, it *could* represent a circular proxy - in that any URL that starts with "http" will get changed to "http://localhost" - which, if it gets processed again - will try to change "http://localhost" into "http://localhost://localhost", etc.

My guess is that this is *not* happening, as otherwise there would be other problems being reported - but still, it's a questionable configuration.

--fish
Senior Consultant, Quotient Inc.
http://www.quotient-inc.com

Migrateduser

Thanks. I will send this thread to the server admin. I imagine this may have been implemented at the suggestion of IW support.

Migrateduser

> Can you give some insight into the pattern you're seeing in memory usage

iwproxy had comsumed 2 MB after service restart yesterday, 12 MB today.

Migrateduser

Proxy is now at 18.5 MB. Is that normal? If it continues to kreep it will need to be restarted or rebooted periodically. Is that really best practice?

Adam Stoller

I'd say "no" it's not normal. If you can - try commenting out that fullproxy_redirect and see (a) what problems occur for people using the system and (b) whether it "fixes" the growth-rate of iwproxy.

If problems occur for the users but it does fix the growth-problem - you've at least narrowed down the source of the problem and can work on trying to find a better solution (like as per my previous post).

If the growth-problem persists and there's no problem for the users - keep the line commented out (with a comment above indicating when you commented it out and why) - and search for another source of the problem. (if problems exist for the users and the growth-problem persists - put back the commented-out line and search for an alternate source).

You may want to contact Support if you haven't already.

--fish
Senior Consultant, Quotient Inc.
http://www.quotient-inc.com

Migrateduser

35MB this morning - that's quite a jump for what was probably no real overnight activity. I don't control the config or the proxy so I sent the thread to the admin.

Thanks,

-John

Migrateduser

Hmmm, down to 12 MB today, and nobody claims to have restarted the service. Does that make any sense?