[Zope-dev] Bad Marshal Data
I've asked this before, about a year ago, but got no response. Does anyone else ever see 'bad marshal data' errors sometimes? What are they a symptom of? It sometimes happens in my products, other times in Zope. It always occurs on import statements. I usually manage to fix it by playing around with line spacing, etc, in the module being imported. Deleting *pyc files does not help. Here's an example: Traceback (most recent call last): File /usr/local/Zope-2.5.1b1-src/lib/python/OFS/Application.py, line 531, in import_product product=__import__(pname, global_dict, global_dict, silly) ValueError: bad marshal data seb ___ Zope-Dev maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope )
Re: [Fwd: [Zope-dev] Re: more on the segfault saga]
Martijn Jacobs wrote: Hello Leo! Tell us if your segfaults go away and whether Zope performance is impacted by it. The segfault is AWAY! I have an uptime of 2 hours now, it has never been so long since it's in use! I don't notice any performance differences, but I also have to say that I can't measure it really. The more important is that it doesn't crash anymore! Is this a *final* solution for the crashing? Or is this just a temporary solution to make sure where the problem is? BTW : Thank Matt for the patch! regards, Martijn. It's going to be a temporary work around, but we're going to build Zope 2.4.4 final and Zope 2.5.1 final around it; I'll be building beta releases this Friday with the patch included. -- Matt Kromer Zope Corporation http://www.zope.com/ ___ Zope-Dev maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] OpenSSH configuration between ZEO clients storage server
Itamar Shtull-Trauring wrote: Toby Dickenson wrote: but ssh port forwarding is only one layer of TCP. ssh port forwarding is good. I'mp pretty sure it's TCP over TCP (ssh protocol does multiplexing). SSL OTOH is not TCP over TCP. I think you are mistaken. :-) The site you referred to says that TCP over TCP is a bad idea because TCP always assumes an unreliable transport, and stacking the compensation results in excessive retries and a far less reliable connection. I agree, this is likely to be a problem. But you don't need reliability compensation to multiplex. SSH assumes the transport layer is reliable. So an SSH tunnel and an SSL tunnel are virtually synonymous. Now, if ZEO accepted SSL connections directly, it would be quite secure and easy to use. But an out-of-process SSL tunnel would suffer from exactly the same difficulties an SSH tunnel does. The Python standard library has good support for an SSL client but not for an SSL server. Shane ___ Zope-Dev maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] OpenSSH configuration between ZEO clients storageserver
Shane Hathaway wrote: But you don't need reliability compensation to multiplex. SSH assumes the transport layer is reliable. So an SSH tunnel and an SSL tunnel are virtually synonymous. They're not really, but yeah, I was wrong :) A SSL forwarder takes a packet, and then sends over another TCP connection that happens to be encrypted. It's basically a TCP port forwarder. Multiple client connections through the local SSL forwarder (e.g. stunnel) will run over *multiple* outgoing streams: client -- stunnel on localhost --- stunnel on www.foo.com server on www.foo.com SSH OTOH uses a multiplexing protocol, where multiple streams of data run over the same connection, and each have their own flow control. Multiple client connections over a SSH port forwarder will run over the same *single* stream. A third thing is TCP over TCP, i.e. PPP over SSH or SSL, which functions as a full network connection. The Python standard library has good support for an SSL client but not for an SSL server. pyOpenSSL is good for both servers and clients. Could be better, but it's easy to start using it, especially for select() based servers (its thread-safety is iffy). Twisted's SSL stuff uses it, and latest version of pyOpenSSL may have asyncore example. ___ Zope-Dev maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] ZClass Constructor Cleanup for 2.6
Casey Duncan wrote: Does anyone see a problem with changing the default generated constructor method for ZClasses to a python script in Zope 2.6? I think we are encouraging really bad style by keeping this in DTML, since it is purely business logic. Thoughts? I will, of course, implement this if nobody objects. +2 (I'm taking Andreas' vote because he says 'zclasses sux' and 'zclasses must die', so he's biased) ___ Zope-Dev maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] ZClass Constructor Cleanup for 2.6
On Wed, 2002-03-27 at 17:31, Casey Duncan wrote: I don't think this is a big enough change to warrant a real proposal, so I'll shoot this out here: Does anyone see a problem with changing the default generated constructor method for ZClasses to a python script in Zope 2.6? I think we are encouraging really bad style by keeping this in DTML, since it is purely business logic. Thoughts? I will, of course, implement this if nobody objects. +5 ZClasses have been overlooked enough. Nobody has even bothered putting a ZCatalogPathAware available for them... -- Ideas don't stay in some minds very long because they don't like solitary confinement. ___ Zope-Dev maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] Bad Marshal Data
I haven't seen anything like that. What I have seen, an which scares the heck out of me everytime I see it is UnpicklingErrror, which, in my case, usually happens when the _p_jar dictionary doesn't have an object that other parts of zope swear they exist. The solution to that usually involves trimming the last transaction(s) off of the ZODB or, when that's not an option, going in thru the monitor port and forcibly deleting the offending objects, where finding the offending objects is an adventure in itself. The problem you're talking about sounds a lot like the ZODB reference to a ceirtan product got corrupted somehow, and forcing the reload of this product (by changing it's syntax slightly) makes the problem go away. On Wed, 2002-03-27 at 08:19, seb bacon wrote: I've asked this before, about a year ago, but got no response. Does anyone else ever see 'bad marshal data' errors sometimes? What are they a symptom of? It sometimes happens in my products, other times in Zope. It always occurs on import statements. I usually manage to fix it by playing around with line spacing, etc, in the module being imported. Deleting *pyc files does not help. Here's an example: Traceback (most recent call last): File /usr/local/Zope-2.5.1b1-src/lib/python/OFS/Application.py, line 531, in import_product product=__import__(pname, global_dict, global_dict, silly) ValueError: bad marshal data seb ___ Zope-Dev maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope ) -- Ideas don't stay in some minds very long because they don't like solitary confinement. ___ Zope-Dev maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] ZClass Constructor Cleanup for 2.6
Good point, I'll add that to my list. -Casey Leonardo Rochael Almeida wrote: On Wed, 2002-03-27 at 17:31, Casey Duncan wrote: I don't think this is a big enough change to warrant a real proposal, so I'll shoot this out here: Does anyone see a problem with changing the default generated constructor method for ZClasses to a python script in Zope 2.6? I think we are encouraging really bad style by keeping this in DTML, since it is purely business logic. Thoughts? I will, of course, implement this if nobody objects. +5 ZClasses have been overlooked enough. Nobody has even bothered putting a ZCatalogPathAware available for them... ___ Zope-Dev maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope )
ZCatalogPathAware (was Re: [Zope-dev] ZClass Constructor Cleanup for 2.6)
Leonardo Rochael Almeida wrote: ZClasses have been overlooked enough. Nobody has even bothered putting a ZCatalogPathAware available for them... I tried do to this quickly for someone on IRC once. The problem is that the mixin is identically named for CatalogAwareness and CatalogPathAwareness, and as such they show up looking exactly the same in the ZClass mixin list. :-) There's probably a way around this that wasn't immediately obvious at the time. ___ Zope-Dev maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope )
[Zope-dev] Crashing problem? Python workaround?!
Hey fellow Zopistas: If you're still experiencing crashing problems with Python 2.1.2, and Zope, please consider trying the following test: From a clean Python build tree, modify Include/object.h line 587. #define PyTrash_UNWIND_LEVEL 50 make this read #define PyTrash_UNWIND_LEVEL 500 Rebuild Python, and Zope, with the new patched Python. Test it. If this makes your problem go away, please post here to the list. Technical details: this patch prevents Python from storing objects in an internal trashcan of objects which need to be deleted, but aren't -- a way to avoid C stack recursion. The trashcan code is not safe from the normal garbage collector; it steals a few fields in the object that the GC depends on (the reference count and the object type field, in fact) and so items in the trashcan will cause the garbage collector to crash with a segmentation fault. Objects only go into the trashcan when a recursive delete is detected, with a depth defined by PyTrash_UNWIND_LEVEL. Bumping this number up to a large number has the effect of disabling the trashcan module, at the expense of the C stack (because all of those recursive deletes CAN happen then.) Pythonlabs found this today after much assistance from the folks on #zope on the irc.openprojects.net system; particularly, Leonardo Rochael Almeida's assistance has been extremely valuable in isolating this problem. This is an intermediate workaround until Pythonlabs develops an official fix. However, by using this workaround, you do not need the specially modified ZServerPublisher module I coded. -- Matt Kromer Zope Corporation http://www.zope.com/ ___ Zope-Dev maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope )
re: [Zope-dev] Crashing problem? Python workaround?!
Hi Matt, Leo Pythonlabs found this today after much assistance from the folks on #zope on the irc.openprojects.net system; particularly, Leonardo Rochael Almeida's assistance has been extremely valuable in isolating this problem. And jus to say it again : thank you all again!! You've made a customer very happy again! This is an intermediate workaround until Pythonlabs develops an official fix. However, by using this workaround, you do not need the specially modified ZServerPublisher module I coded. Will this patch be included in the next Zope binary releases? I could build a zope for on a particular environment, but it would be alot easier to upgrade all customers with the new binary release. Anyway : In other words : if this is added in the official Python release, the bug is cleanly resolved? -- martijn jacobs east site ___ Zope-Dev maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope )