Does ZSyncer have any problem with Zope/ZEO systems? I have some anecdotal evidence of problems as we migrate to a multiheaded Zope/ZEO systems architecture. Is there any difficult using ZSyncer between a Zope and a Zope/ZEO combination. Are there problems syncing between Zope 2.6.X and Zope 2.7.X systems? All the systems run ZSyncer 0.5.1.
I sent roughly this email directly to Shane Hathaway, since I hadn't asked for any help on Zope, yet, so didn't know where to look. Maybe this was a better avenue? If you read this list, Shane (or Mr. Hathaway, if you prefer), respond on whichever seems more appropriate, I guess. I don't know, let me know if I'm missing some etiquette point. Anyways, I'm using Ape 1.0 set-up in the root of Zope 2.7.2 install, based on the example fs config, and that is working fine. I was having a problem where some change was making that Zope instance not even get far enough in booting to write to the event.log (set to level all). I narrowed it down to adding a local role on the security tab of the root object. I revert the .properties file and restart - it works. I make the change again and restart - it doesn't. I diff the file, and the only difference is the define-role line. The problem is I have no idea where to start even, since no errors are thrown where I know to find them. Zope just doesn't respond at all when I try to restart it, and writes nothing to the log. If anyone could tell me what I can do to at least track down the problem, if not fix it, it would be much appreciated. Thanks, Tim
Tim, You may want to run Zope in debug mode-- bin/runzope should do it. Also, be sure you have the event.log configured and enable. -d On Wed, 25 May 2005, Tim Connor wrote:
I sent roughly this email directly to Shane Hathaway, since I hadn't asked for any help on Zope, yet, so didn't know where to look. Maybe this was a better avenue? If you read this list, Shane (or Mr. Hathaway, if you prefer), respond on whichever seems more appropriate, I guess. I don't know, let me know if I'm missing some etiquette point.
Anyways, I'm using Ape 1.0 set-up in the root of Zope 2.7.2 install, based on the example fs config, and that is working fine. I was having a problem where some change was making that Zope instance not even get far enough in booting to write to the event.log (set to level all). I narrowed it down to adding a local role on the security tab of the root object. I revert the .properties file and restart - it works. I make the change again and restart - it doesn't. I diff the file, and the only difference is the define-role line. The problem is I have no idea where to start even, since no errors are thrown where I know to find them. Zope just doesn't respond at all when I try to restart it, and writes nothing to the log.
If anyone could tell me what I can do to at least track down the problem, if not fix it, it would be much appreciated.
Thanks, Tim _______________________________________________ Zope maillist - Zope@zope.org http://mail.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope-dev )
-- Dennis Allison * Computer Systems Laboratory * Gates 227 * Stanford University * Stanford CA 94305 * (650) 723-9213 * (650) 723-0033 fax * allison@shasta.stanford.edu * allison@sumeru.stanford.edu
Thanks, I feel like an idiot - running it as a daemon would explain why I wasn't getting as much feedback. I still don't quite get why it wasn't writing to the log though, since it is normally doing that fine. I've had that frustration with Zope before though - some errors seem to kill it in such a way it never touches the log file, even when set to all. Shane responded, so I'll see if it is an obvious Ape problem with roles on root, and then either follow-up here with a solution/more info, in case it helps someone else, or hoping for more help, if not. Well I'll throw the trace in here, in case anyone has any bright ideas, but not expect anybody to put to much thought in until we see what he has to say. ;) ------ 2005-05-25T01:25:06 INFO(0) ZServer HTTP server started at Wed May 25 01:25:06 2005 Hostname: gibson Port: 9280 ------ 2005-05-25T01:25:06 INFO(0) ZServer FTP server started at Wed May 25 01:25:06 2005 Hostname: gibson Port: 9221 ------ 2005-05-25T01:25:06 INFO(0) Zope Set effective user to "zope-is" ------ 2005-05-25T01:25:08 DEBUG(-200) ApeStorage create storage ApeStorage: /var/lib/zope/zope-is/spicerack/var/root/ Traceback (most recent call last): File "/usr/lib/zope-2.7.2/lib/python/Zope/Startup/run.py", line 50, in ? run() File "/usr/lib/zope-2.7.2/lib/python/Zope/Startup/run.py", line 19, in run start_zope(opts.configroot) File "/usr/lib/zope-2.7.2/lib/python/Zope/Startup/__init__.py", line 51, in start_zope starter.startZope() File "/usr/lib/zope-2.7.2/lib/python/Zope/Startup/__init__.py", line 230, in startZope Zope.startup() File "/usr/lib/zope-2.7.2/lib/python/Zope/__init__.py", line 47, in startup _startup() File "/usr/lib/zope-2.7.2/lib/python/Zope/App/startup.py", line 96, in startup OFS.Application.initialize(application) File "/usr/lib/zope-2.7.2/lib/python/OFS/Application.py", line 278, in initialize initializer.initialize() File "/usr/lib/zope-2.7.2/lib/python/OFS/Application.py", line 301, in initialize self.install_required_roles() File "/usr/lib/zope-2.7.2/lib/python/OFS/Application.py", line 457, in install_required_roles app.__ac_roles__=app.__ac_roles__ + ('Owner',) TypeError: can only concatenate list (not "tuple") to list Thanks, Tim Tim Dennis Allison wrote:
Tim, You may want to run Zope in debug mode-- bin/runzope should do it. Also, be sure you have the event.log configured and enable. -d
On Wed, 25 May 2005, Tim Connor wrote:
I sent roughly this email directly to Shane Hathaway, since I hadn't asked for any help on Zope, yet, so didn't know where to look. Maybe this was a better avenue? If you read this list, Shane (or Mr. Hathaway, if you prefer), respond on whichever seems more appropriate, I guess. I don't know, let me know if I'm missing some etiquette point.
Anyways, I'm using Ape 1.0 set-up in the root of Zope 2.7.2 install, based on the example fs config, and that is working fine. I was having a problem where some change was making that Zope instance not even get far enough in booting to write to the event.log (set to level all). I narrowed it down to adding a local role on the security tab of the root object. I revert the .properties file and restart - it works. I make the change again and restart - it doesn't. I diff the file, and the only difference is the define-role line. The problem is I have no idea where to start even, since no errors are thrown where I know to find them. Zope just doesn't respond at all when I try to restart it, and writes nothing to the log.
If anyone could tell me what I can do to at least track down the problem, if not fix it, it would be much appreciated.
Thanks, Tim _______________________________________________ Zope maillist - Zope@zope.org http://mail.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope-dev )
Thanks Dennis, Between your and Shane's reply, I got it fixed: I just raised an error with the value of app.__ac_roles__, at that point in lib/python/OFS/Application.py and determined that ONLY the role I had added was being created. So going in and manually adding those starting default roles to the .properties file fixed it (running it from the console 2 days ago sure would have saved me some sleep :/). I have no idea if that is a problem with how I moved an existing project into a root install of Ape, or if it is a more generic one, but on my set-up Ape definitely only puts things in that [security] section of the .properties files when you make a change. (For example, just hitting save on the Security tab adds all those per role settings, but not the roles themselves). On everything aside from root that works better, but obviously if you ever get to the point I was at where there is only one role defined on start-up, it kill Zope, badly. Technically that might be a Zope error also - if there is only one role, it hits a type mismatch. But people shouldn't ever be getting there anyways, I suppose. Tim Dennis Allison wrote:
Tim, You may want to run Zope in debug mode-- bin/runzope should do it. Also, be sure you have the event.log configured and enable. -d
On Wed, 25 May 2005, Tim Connor wrote:
I sent roughly this email directly to Shane Hathaway, since I hadn't asked for any help on Zope, yet, so didn't know where to look. Maybe this was a better avenue? If you read this list, Shane (or Mr. Hathaway, if you prefer), respond on whichever seems more appropriate, I guess. I don't know, let me know if I'm missing some etiquette point.
Anyways, I'm using Ape 1.0 set-up in the root of Zope 2.7.2 install, based on the example fs config, and that is working fine. I was having a problem where some change was making that Zope instance not even get far enough in booting to write to the event.log (set to level all). I narrowed it down to adding a local role on the security tab of the root object. I revert the .properties file and restart - it works. I make the change again and restart - it doesn't. I diff the file, and the only difference is the define-role line. The problem is I have no idea where to start even, since no errors are thrown where I know to find them. Zope just doesn't respond at all when I try to restart it, and writes nothing to the log.
If anyone could tell me what I can do to at least track down the problem, if not fix it, it would be much appreciated.
Thanks, Tim _______________________________________________ Zope maillist - Zope@zope.org http://mail.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope-dev )
I spoke too soon. I don't think it's Zope error on that type mismatch (or if it is one, it's not that one I thought). It still happens if there are multiple roles defined. As long as it hits the code adding the non-existing Owner role* the start-up croaks. So either something about Ape is causing the type mis-match or the Zope code has a bug and just usually doesn't hit that branch (which is possible, as I suspect most people don't manage to delete the Owner user). I don't know python, very well, so I don't have much of an idea. I tried changing the syntax on the ('Owner',) a couple different ways to get around the type mis-match, and then I got things started, but without any of the default users. Now it seems even more like there is an Ape bug involved (or at least a bug in getting it working on the root). Whenever I save the security tab on the root folder, it rewrites the .properties file, without the default roles defined, which puts Zope in an unstartable place. But then if I tweak Zope to make it start the default roles aren't defined. It seems I need to explicitly define those roles in the root .properties, but even if I do that manually then I can never touch the security tab (not sure if changing any other property on root causes the same problem). Tim * Since the traceback isn't in this set of replies, heres the pertinent code: File "/usr/lib/zope-2.7.2/lib/python/OFS/Application.py", line 457, in install_required_roles app.__ac_roles__=app.__ac_roles__ + ('Owner',) TypeError: can only concatenate list (not "tuple") to list Tim Connor wrote:
Thanks Dennis,
Between your and Shane's reply, I got it fixed:
I just raised an error with the value of app.__ac_roles__, at that point in lib/python/OFS/Application.py and determined that ONLY the role I had added was being created. So going in and manually adding those starting default roles to the .properties file fixed it (running it from the console 2 days ago sure would have saved me some sleep :/). I have no idea if that is a problem with how I moved an existing project into a root install of Ape, or if it is a more generic one, but on my set-up Ape definitely only puts things in that [security] section of the .properties files when you make a change. (For example, just hitting save on the Security tab adds all those per role settings, but not the roles themselves). On everything aside from root that works better, but obviously if you ever get to the point I was at where there is only one role defined on start-up, it kill Zope, badly.
Technically that might be a Zope error also - if there is only one role, it hits a type mismatch. But people shouldn't ever be getting there anyways, I suppose.
Tim
Dennis Allison wrote:
Tim, You may want to run Zope in debug mode-- bin/runzope should do it. Also, be sure you have the event.log configured and enable. -d
On Wed, 25 May 2005, Tim Connor wrote:
I sent roughly this email directly to Shane Hathaway, since I hadn't asked for any help on Zope, yet, so didn't know where to look. Maybe this was a better avenue? If you read this list, Shane (or Mr. Hathaway, if you prefer), respond on whichever seems more appropriate, I guess. I don't know, let me know if I'm missing some etiquette point.
Anyways, I'm using Ape 1.0 set-up in the root of Zope 2.7.2 install, based on the example fs config, and that is working fine. I was having a problem where some change was making that Zope instance not even get far enough in booting to write to the event.log (set to level all). I narrowed it down to adding a local role on the security tab of the root object. I revert the .properties file and restart - it works. I make the change again and restart - it doesn't. I diff the file, and the only difference is the define-role line. The problem is I have no idea where to start even, since no errors are thrown where I know to find them. Zope just doesn't respond at all when I try to restart it, and writes nothing to the log.
If anyone could tell me what I can do to at least track down the problem, if not fix it, it would be much appreciated.
Thanks, Tim _______________________________________________ Zope maillist - Zope@zope.org http://mail.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope-dev )
_______________________________________________ Zope maillist - Zope@zope.org http://mail.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope-dev )
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Dennis Allison wrote:
Does ZSyncer have any problem with Zope/ZEO systems? I have some anecdotal evidence of problems as we migrate to a multiheaded Zope/ZEO systems architecture.
Is there any difficult using ZSyncer between a Zope and a Zope/ZEO combination.
There shouldn't be any issues there: ZSyncer runs at the "application" level, not the "database" level.
Are there problems syncing between Zope 2.6.X and Zope 2.7.X systems?
*That* might be problematic, although I can't think of a case offhand.
All the systems run ZSyncer 0.5.1.
Tres. - -- =================================================================== Tres Seaver tseaver@palladion.com Palladion Software "Excellence by Design" http://palladion.com -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.5 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFClGJy+gerLs4ltQ4RAqZzAKCC1aj/6x9z5DEBcwtoY66dL33DqACgq8CY AmaeKQ6cb8R0hDyqIaLFfUw= =Hyq6 -----END PGP SIGNATURE-----
Furher investigation has shown that the failing ZSyncer target had a problem with its authentication code and that was likely the cause of the failure. We are continuing to test, but it appears everything is working. We'll continue to test but I am pretty sure this was not a ZSyncer problem. Thanks to all. -d On Wed, 25 May 2005, Tres Seaver wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Dennis Allison wrote:
Does ZSyncer have any problem with Zope/ZEO systems? I have some anecdotal evidence of problems as we migrate to a multiheaded Zope/ZEO systems architecture.
Is there any difficult using ZSyncer between a Zope and a Zope/ZEO combination.
There shouldn't be any issues there: ZSyncer runs at the "application" level, not the "database" level.
Are there problems syncing between Zope 2.6.X and Zope 2.7.X systems?
*That* might be problematic, although I can't think of a case offhand.
All the systems run ZSyncer 0.5.1.
Tres. - -- =================================================================== Tres Seaver tseaver@palladion.com Palladion Software "Excellence by Design" http://palladion.com -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.5 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
iD8DBQFClGJy+gerLs4ltQ4RAqZzAKCC1aj/6x9z5DEBcwtoY66dL33DqACgq8CY AmaeKQ6cb8R0hDyqIaLFfUw= =Hyq6 -----END PGP SIGNATURE-----
-- Dennis Allison * Computer Systems Laboratory * Gates 227 * Stanford University * Stanford CA 94305 * (650) 723-9213 * (650) 723-0033 fax * allison@shasta.stanford.edu * allison@sumeru.stanford.edu
On Wed, May 25, 2005 at 09:15:32AM -0700, Dennis Allison wrote:
Furher investigation has shown that the failing ZSyncer target had a problem with its authentication code and that was likely the cause of the failure. We are continuing to test, but it appears everything is working. We'll continue to test but I am pretty sure this was not a ZSyncer problem.
You might be interested to know that I made substantial improvements to the error reporting and logging for ZSyncer 0.6. Problems like this are generally easier to diagnose now. -PW -- Paul Winkler http://www.slinkp.com
I noted that and will upgrade in this round. It's the old problem of having live systems, distributed development, and all. The usual groundrules are: only transparent changes, no risk of errors, no disruption of 24x7 use, etc. On Wed, 25 May 2005, Paul Winkler wrote:
On Wed, May 25, 2005 at 09:15:32AM -0700, Dennis Allison wrote:
Furher investigation has shown that the failing ZSyncer target had a problem with its authentication code and that was likely the cause of the failure. We are continuing to test, but it appears everything is working. We'll continue to test but I am pretty sure this was not a ZSyncer problem.
You might be interested to know that I made substantial improvements to the error reporting and logging for ZSyncer 0.6. Problems like this are generally easier to diagnose now.
-PW
-- Dennis Allison * Computer Systems Laboratory * Gates 227 * Stanford University * Stanford CA 94305 * (650) 723-9213 * (650) 723-0033 fax * allison@shasta.stanford.edu * allison@sumeru.stanford.edu
On Tue, May 24, 2005 at 11:29:02PM -0700, Dennis Allison wrote:
Does ZSyncer have any problem with Zope/ZEO systems? I have some anecdotal evidence of problems as we migrate to a multiheaded Zope/ZEO systems architecture.
I'd be curious to hear those anecdotes. Where I work, we use zsyncer daily and we *always* use ZEO. The only potential issue I could see is that ZEO adds some network overhead to ZODB writes, increasing the chances of a conflict error (or, in recent versions of zsyncer, a timeout).
Is there any difficult using ZSyncer between a Zope and a Zope/ZEO combination.
Are there problems syncing between Zope 2.6.X and Zope 2.7.X systems?
We synced from 2.7.X to 2.6.X for quite a while with no observed problems. (Our development systems are often more current than our production systems.) Never tried the other way around AFAIK.
All the systems run ZSyncer 0.5.1.
... which, just so you know, is now different enough from the current codebase that if you had any problems with it, I'd be unable to give you any advice except to upgrade :-) -- Paul Winkler http://www.slinkp.com
On Wed, May 25, 2005 at 10:14:23AM -0400, Paul Winkler wrote:
We synced from 2.7.X to 2.6.X for quite a while with no observed problems.
... which, to be clear, you should take as anecdotal evidence only. I can't guarantee that you will have no problems syncing between mismatched zope versions. -- Paul Winkler http://www.slinkp.com
participants (4)
-
Dennis Allison -
Paul Winkler -
Tim Connor -
Tres Seaver