Note: The mentioned enclosure was removed for transmission to the mailing list. ---------- From: Jim Fulton [SMTP:jim.fulton@digicool.com] <mailto:[SMTP:jim.fulton@digicool.com]> Sent: Friday, February 12, 1999 6:56 PM To: Shaw, Howard Cc: support@digicool.com; <mailto:support@digicool.com;> zope@zope.org <mailto:zope@zope.org> Subject: Re: [Zope] Shattered dreams... "Shaw, Howard" wrote: > (snip) > Have you just started using Zope? Or have you been using > Zope for a while without problems? > > I started using Zope fairly recently, and had only gone so far > as to > make it a mirror of my pre-existing site using fsimport. > > Can you take MySQL out of this mix for a while to see if it has > any > effect? Are there any other products installed that are not > part of the Zope distribution? > > I can say that having just reinstalled the zope binary distro, I > saw no > regular stoppage of the zope server until after I reinstalled > the ZMySQLDA > and connected to the database. I have not manually installed any > other products, > though I did install the external method 'fsimport'. Please try running Zope without the ZMySQLDA and see if that has any effect As in remove the product; or disconnect the database connection? (snip) > It would also be interesting to us to see your database files. > Would you mind sending them to us (as an attached gzipped > tar ball or Zip file)? > > I assume you mean the Data.BBB* files. Actually, the var directory. > Do you want the files > from the current > working setup (which is exhibiting the recurrent loss of Zope with no > messages) > or the non-working Data.BBB, or both? I'm confused. Are you saying that you have a working setup that, asside from having ZopeHTTPServer die, is not having problems? How does this relate to the broken site. What I'm saying, is that I had a working setup; subsequently it developed a periodic death syndrome; then finally cratered completely after this wierd session problem. I presumed a connection, with no evidence, simply because it seems logical to me from the description I've read of the Bobo database, that the Zope server being cut off while accessing the file might leave the database in an uncertain state. The scenario I pictured was one where some portion of the basic database structure was corrupted, but the region internal to the session was not. Work within the session went fine, until the save, where it attempted to access the corrupt portion, failed, rolled back the changes, but was not able to 'undo' the session. This fits your general documentation, which implies that while individual changes can be individually undone, the application of a session can only be undone in the sense of a complete removal, and not of a restoration of the session to its prior state. Subsequent to the session difficulty, I started editing in the open. I only managed to touch four or five files before the database died and then refused to restart, which makes me think that I then manipulated the damaged region in such a way as to cause damage to the structural information, whereas before the damage was to the internals of individual items. Of course, all of this is complete guesswork. I never saw any useful log message other than the final message about a corrupt record. Therefore the fact that my ZopeHTTPServer is periodically dying implies that a similar crash lurks in my future. The reason I was unsure which you wanted, is that my theories imply that this working DB may contain information about the cause of the crash, while the other DB is completely inaccessible, so far as I can tell. Of course, I realize that working more directly with Bobo, one might be able to extract data more effectively than through Zope, which of course has a greater set of dependencies. For starters, please send the var directory containing the non-working Data.bbb files as a gzipped tar file. It is included. > I assume you don't want these > mailed to > the list; Right. Please send to jim@digicool.com <mailto:jim@digicool.com> . (Note that I didn't CC the list in my message to you.) > should I send them to your personal account? Yes. jim@digicool.com <mailto:jim@digicool.com> . > The problems you are reporting are very odd. Needless to say, > we don't see them here, so any info you give would be very > helpfull. > > Please feel free to call me at the number below if you're > interested in brainstorming this. > > I am very interested in determining the cause of and solution > for this problem, especially > since it was, in my opinion, related to the recurrent death of > the ZopeHTTPServer, > which has returned. When I moved the non-working installation to a different directory, and installed a fresh copy of the binary package, I noticed that the first spontaneous crash did not come until after I installed the MySQLDA, even though I spent at least two hours rebuilding the links between a separate portion of the Zope site and my original site. (Using it as a replacement mirror.) The recurrent death returned during my rebuilding of the database using portion of the site. BTW, this is using the Zope internal version of Python with MySQLModule.so compiled from the source obtained with MySQLDA. Did it stop dying? When? Was it related to the use of MySQLDA? > Since I believe it to be only a matter of > time until it dies again, I am > keeping multiple backups of the Data.BBB* files. Are there any > other data files I should > be aware of, for purposes of backup, or do the Data.BBB files > cover it? The Data.bbb files cover it. Jim -- Jim Fulton mailto:jim@digicool.com <mailto:jim@digicool.com> Technical Director (888) 344-4332 Python Powered! Digital Creations http://www.digicool.com <http://www.digicool.com> http://www.python.org <http://www.python.org> Under US Code Title 47, Sec.227(b)(1)(C), Sec.227(a)(2)(B) This email address may not be added to any commercial mail list with out my permission. Violation of my privacy with advertising or SPAM will result in a suit for a MINIMUM of $500 damages/incident, $1500 for repeats.