May 2001 - Zope-Dev - Zope lists

CopySupport bug
by Casey Duncan 03 May '01

03 May '01

There is a subtle bug in the CopySupport module that makes it impossible to copy and paste a folder that contains a ZCatalog with CatalogAware objects in indexed in the catalog also in that folder. The reason is as follows, CopySupport calls a hookable method manage_afterClone (which is hooked by CatalogAware). It does so after a copy is made of the object but BEFORE the copy is mounted in the ZODB. manage_afterClone in CatalogAware tries to reindex the object, and in doing so tries to derive its URL for use as the ZCatalog uid by calling absolute_url on the object. This fails of course because the object as yet has no URL because it is not mounted. The error reported is a KeyError on SERVER_URL. To make a long story somewhat shorter, I have a simple patch for this, but there is still one more subtle bit of behavior to take note of: When the ZCatalog is copied, all of its indexed data gets copied. Then the new copies of the objects index themselves. This means that the copied catalog has twice as many objects in it as the original. This bugged me at first, but it makes sense and I don't see any obvious solution that would be consistent. For those interested, the patch is here: http://classic.zope.org:8080/Collector/2205/view -- | Casey Duncan | Kaivo, Inc. | cduncan(a)kaivo.com `------------------>

1 0

Re: [Zope-dev] ANN: Proposal ContentNegotiation
by Andreas Jung 03 May '01

03 May '01

From: "Toby Dickenson" <tdickenson(a)geminidataloggers.com> To: "'Andreas Jung'" <andreas(a)andreas-jung.com> Sent: Thursday, May 03, 2001 1:04 PM Subject: RE: [Zope-dev] ANN: Proposal ContentNegotiation > Your proposal suggests that published objects should generally return > pre-encoded objects in 8-bit strings. This is the one detail which in my > experience looks *wrong* and is very much dependant on my proposal. Not really - I only say that the internal character set is currently ascii/iso-8859-1. I did not mention that this will be the default character set in the future. Andreas

1 0

Re: [Zope-dev] another Zope hanging..
by Dieter Maurer 03 May '01

03 May '01

Christian Scholz writes: > I also now have some problems with a hanging zope. > > I got something like this in my stupid logfile: > > 2001-05-03T13:44:39 ERROR(200) ZServer uncaptured python exception, closing channel <select-trigger (pipe) at 81d64c8> (exceptions.OSError:[Errno 11] Die Ressource ist zur Zeit nicht verf?gbar [/opt/Zope-2.3.1-src/ZServer/medusa/asyncore.py|poll|83] [/opt/Zope-2.3.1-src/ZServer/medusa/asyncore.py|handle_read_event|335] [/opt/Zope-2.3.1-src/ZServer/medusa/select_trigger.py|handle_read|77] [/opt/Zope-2.3.1-src/ZServer/medusa/asyncore.py|recv|287] [/opt/Zope-2.3.1-src/ZServer/medusa/asyncore.py|recv|475]) What you see here is a traceback pressed into a single line. [/opt/Zope-2.3.1-src/ZServer/medusa/asyncore.py|recv|475] e.g. means: you are in file "/opt/Zope-2.3.1-src/ZServer/medusa/asyncore.py" line "475" in function "recv". Dieter

1 0

catalog and versions
by Stefane Fermigier 03 May '01

03 May '01

It seems to me that when I create a CatalogAware objet in a version, it doesn't get into the catalog after the version is committed (in fact, it doesn't event get into the catalog inside the version). Am I right? Is this considered normal behavio(u)r? Any idea for a workaround? Regards, S. -- Stéfane Fermigier, Tel: +33 (0)6 63 04 12 77 (mobile). http://nuxeo.com/ & http://portalux.com/ & http://aful.org/ "Amazon: we patent the dot in .com"

1 0

ANN: Proposal ContentNegotiation
by Andreas Jung 03 May '01

03 May '01

Hi folks, feel free to review and comment the proposal for extending the content negotiation capabilities of Zope: http://dev.zope.org/Wikis/DevSite/Proposals/HTTPContentNegotiation Cheers, Andreas Digital Creations

2 2

RE: [Zope-dev] Zope Server hanging :-(
by Tim McLaughlin 03 May '01

03 May '01

I forgot to mention... our neither our z2.log or stupid_log show anything upon freezing. afterwards the stupid_log shows the failed transaction cleanup, but that's it. Tim -----Original Message----- From: Tim McLaughlin Sent: Friday, April 27, 2001 12:41 PM To: 'zope-dev(a)zope.org'; 'chrisw(a)nipltd.com' Subject: [Zope-dev] Zope Server hanging :-( Chris (and all), We seem to be having similar crashing issues. And interestingly enough, they don't seem to show up on any of our servers except the 2.3.0 upgraded to 2.3.1. 1. Hangs on anything from Pythonscript or dtml 2. top shows no significant mem or cpu usage 3. threads are all unresponsive 4. restart usually yields a .trX file (an aborted transaction I suppose) Any ideas? Maybe it was something "fixed" in 2.3.1? The causes seem to be somewhat different except that they all modify the ZODB.... Tim

3 14

RE: [Zope-dev] Zope Server hanging :-(
by Tim McLaughlin 03 May '01

03 May '01

Nice info. This could be useful for many, many things. you oughtta put this in a how-to (if you haven't already). thanks Chris. I'll use it if we have any more issues. Currently, as I told ChrisW, the issues have stopped. But I'm still keeping my eye on it. -----Original Message----- From: Chris McDonough [mailto:chrism@digicool.com] Sent: Wednesday, May 02, 2001 12:00 PM To: Chris Withers Cc: Tim McLaughlin; zope-dev(a)zope.org Subject: Re: [Zope-dev] Zope Server hanging :-( One of the more important bits is the "active" count when using the --detailed option to requestprofiler. This tells you how many *other* requests were "unfinished" at the end of a particular request. So for instance, if you run requestprofiler like this: python requestprofiler.py log.file.name --detailed And you see a profile line like (header added for reference): start win wout wend tot S osize code act url 2001-04-19T01:19:42 0 12 0 12 E 26024 200 5 /Catalog/checkrep?P This means that: - this request started at 2001-04-19 1:19:42 UTC ("start") - zserver waited 0 secs for input from the client ("win") - zserver waited 12 secs for output from the publisher ("wend") - the total time from start to end for the request was 12 secs ("tot") - the request reached the end ("E") stage ("S", one of "E", "A", "I", "B") - the output size in bytes was 26024 ("osize") - the http code was 200 ("code") - the number of pending requests at the end of this request was 5 ("active") - the URL of the request ("url") Hangs are also interesting. Use detailed reporting to sort by end stage: requestprofiler.py your.log.file --detailed --sort=endstage Any requests that don't show a stage of "E" for their endstage were "hung" requests. You can also use cumulative reporting to sort by "hangs": requestprofiler.py your.log.file --cumulative --sort=hangs This will show the methods that "hung" most frequently. Much of the time, you can find problems this way by seeing methods that enter an infinte loop or cause a segfault. Total time is also interesting: requestprofiler.py your.log.file --detailed --sort=total Shows you individual calls to methods that took a long time. - AND - requestprofiler.py your.log.file --cumulative --sort=total Shows you where most of the request time is spent. The goal is to indentify methods or requests that take a long time, don't return, or are accessed frequently. If you can match this data up with specific problems you've experienced (possibly by way of times/dates), it becomes easier to form postulations about what is happening to Zope under load. Then investigate those postulations by refactoring code or by doing more detailed logging in the places where you think the problems are. - C > Hmmm, not really sure what I should be looking for :-S > > What ways should I look to run it and what should I do with the output? > > cheers, > > Chris >

4 3

ANNOUNCE Berkeley storage 1.0 beta 2 for Zope/ZODB
by barry＠digicool.com 02 May '01

02 May '01

Hello Zope and ZODB users! The folks at Digital Creations, PythonLabs bring to you the second beta release of a Berkeley DB based storage for ZODB. Information about the Berkeley storage, can be found on the Berkeley storage Wiki at http://www.zope.org/Wikis/ZODB/BerkeleyStorage and other ZODB developments can be found on the ZODB Wiki page: http://www.zope.org/Wikis/ZODB/FrontPage This release of the Berkeley storage package contains the Packless.py implementation from 1.0beta1 which provides a non-versioning, non-undoable storage that uses reference counting to eliminate the need for explicit packs (except in the face of object cycles). The 1.0beta2 release adds two new storages: Full.py which is a complete versioning and undoable storage, and Minimal.py which is a rewritten non-versioning, non-undoable storage that will eventually replace Packless.py. Also included is an extensive PyUnit-based test suite. You must install Sleepycat's Berkeley DB 3.2.9 and Robin Dunn's PyBSDDB wrapper package. Berkeley storage 1.0beta2 has been tested with Python 2.0 and 2.1; it may work with Python 1.5.2 but no decision yet has been made whether that will be officially supported. Berkeley storage 1.0beta2 can be installed as a Python distutils package, and works well with StandaloneZODB distribution: http://www.zope.org/Wikis/ZODB/StandaloneZODB It, of course, can also be used with Zope, however please read the release notes concerning undo with the Full.py storage (there's a new non-destructive undo mechanism called Transactional Undo, which is not compatible with Zope 2.3.1 -- Full.py should work with Zope 2.3.1 but you will not be able to undo). Future development will include the integration of cyclic garbage detection with Full.py, and the addition of reference counting and cyclic gc to Minimal.py. See the ZODB Wiki for other developments. The release is available here: http://www.zope.org/Products/bsddb3Storage Discussions should take place on the zodb-dev(a)zope.org mailing list. Enjoy, -Barry Barry A. Warsaw PythonLabs at Digital Creations barry(a)digicool.com

1 0

Bug in dtml-in ????
by Andre Schubert 02 May '01

02 May '01

Hi, I have found a error/bug in the dtml-in tag. I have an SQL-Query that returns traffic-data grouped by month and year. Th SQL-query returns 5 rows of data. I want a page to show only one month. If i use: <dtml-in foo size=1 orphan=1 start=query_start> <dtml-if sequence-start> <dtml-if next-sequence> ..... </dtml-if> </dtml-if> I got the following result. The next-statement returns the url to the next row of data( the second row ). The context is the current row of data ( first row ). If i click on the link to the next row( query_start=2 ) i got the the second row of data, but i don't got a link to the third row of data, i only got a link back to the first. If i set size=2 everything works fine. But when size=1 the the error occurs, i can only switch between the first to rows of data. I have testet it with different SQL-query's on Zope-2.2.5 on Redhat-Immunix-6.2 Please help Thanks as

1 1

Waagh! More fun & games with the CST 0.8 Updater
by Chris Withers 02 May '01

02 May '01

It's running and working okay, but every so often, when running it, I get: 2001-04-27T09:15:23 PANIC(300) ZODB A storage error occurred in the last phase o f a two-phase commit. This shouldn't happen. The application may be in a hosed state, so transactions will not be allowed to commit until the site/storage is r eset by a restart. Traceback (innermost last): File E:\Zope\23CF7D~1.2B2\lib\python\ZODB\Transaction.py, line 335, in commit File E:\Zope\23CF7D~1.2B2\lib\python\ZODB\Connection.py, line 657, in tpc_fini sh File E:\Zope\23CF7D~1.2B2\lib\python\ZODB\BaseStorage.py, line 219, in tpc_fin ish (Object: SessionStorage) error: release unlocked lock The relevant bit of code is: path = self.getSessionDataContainerPath() timeout_mins = self.getInternalDCTimeoutMins() onstart = self.getOnStartPath() onend = self.getOnEndPath() container = self.aq_parent ob = SessionDataManager(id, path, title, timeout_mins, onstart, onend) container.manage_delObjects(ids=[id]) container._setObject(id, ob) Any ideas? cheers, Chris

2 10