Forum OpenACS Development: AOLServer 4.0 Support in 5.0?

Collapse
Posted by Mark Aufflick on
As mentioned again in a recent post (http://openacs.org/forums/message-view?message_id=144504), AOLServer 4.0 is now fully released, and will make everyones life a lot better.

I don't rememebr seeing the outcome of il8n and charset decisions for aol4. The last I heard, the charset implementation was different from the aD patches, but our implementation was being considered.

A google search of openacs.org shows a bunch of fixed bugs for aolserver 4, but I can't find any references to the il8n issues.

Can anyone shed light on how well aol4 is supported vs. the aD patched aol3.3, and perhaps we should TIP moving to aol4 as the oficially supported server for 5.0 since it would be a good time to make such a major change.

Collapse
Posted by Jamie Rasmussen on
The AOLserver 4.0 i18n work was inspired by/based on the AD API, so there shouldn't be many problems.  In the 4.0 draft documentation - http://aolserver.com/40drafts/tclapi/ - you'll see ns_charsets, ns_encodingforcharset, ns_cookiecharset, etc.  No one has written documentation for these yet, but perhaps someone from the OpenACS community will find the time to contribute it.
Collapse
Posted by Jeff Davis on
Mark, In terms of support by openacs.org I would say there are a number of us who have been using 4.0 regularly through the beta period so there is plenty of expertise about how to get it to work (and I have not encountered any problems with it since beta10 I think).

I believe internally AOL has cut over to 4.0 (digitalcity.com definitely has, not sure about other sites) so I am pretty confident it has gotten a good workout and is production ready.

Yes, I think we should switch to 4.0 as the recommended version although I am not sure we can get the install docs revised in time for 5.0 so it might happen with 5.1.

Collapse
Posted by Malte Sussdorff on
I will rewrite the install docs for 5.0 and AOLserver 4.0 til end of the week unless someone else is faster.
Collapse
Posted by Andrew Piskorski on
AFAIK there is still no SSL module that works with AOLserver 4.0. Last we heard on the AOLserver list, Scott Goodwin was still working on it.
Collapse
Posted by Jamie Rasmussen on
It hasn't been released, but the nsopenssl in the SourceForge repository (3.x) works with AOLserver 4.0 for me. I made some minor modifications to compile on Win32 and to fix two bugs - one bug was that the debug trace wasn't respecting an off configuration and one bug was causing blank error messages in the logs. The documentation hasn't been fully updated yet, and the config file settings are quite different than in 2.x, so I'm not suggesting a rush to use it in production, but...

Scott is in the process of a major refactoring and is hoping to release another version in the next few days. If you aren't on the AOLserver mailing list, here's some information on the direction nsopenssl is heading in:
http://www.mail-archive.com/aolserver@listserv.aol.com/msg06022.html
http://www.mail-archive.com/aolserver@listserv.aol.com/msg06044.html

Collapse
Posted by Malte Sussdorff on
ns_openssl has been successfully installed for our Squirrelmail in OpenACS project. Goutham is writing up the documentation as it did not work out that easy. But to come back to the original statment. Yes, we do have AOLserver 4 with ns_openssl running. Should this be enough to warrant the use of AOLserver for OpenACS ?
Collapse
Posted by Joel Aufrecht on
Collapse
9: Two new AOLserver 4.0 bugs (response to 1)
Posted by Andrew Piskorski on
FYI, very recently people have reported some new bugs/problems running AOLserver 4.0:

Restart problems, multiple people reporting having to send a kill signal twice with a pause in between and/or kill -9. Sounds like a problematic and annoying bug but not a show-stopper for most people.

One or two people also seem to be reporting AOLserver 4 respawning infinite times until the box comes to a halt. It is somewhat unclear, but it appears that this is likely due to attempts to remotely exploit an as yet unpatched security bug AOLserver 4.0, and those (presumably unsuccessful) attempts had the effect of a DOS attack. Supposedly this hole is very difficult to exploit successfully, an it is thought that the bug does not exist in AOLserver 3.3+ad13, but AFAIK no one has double checked and verified that.

On a more pleasant note, Zoran recently announced some promising alternative Tcl interp initialization work giving reportedly 1/3 the memory footprint and 50 times faster thread startup than baseline in AOLserver 4.0.