RE: [Zope] ANN: ZEO goes Open Source
Note that what you're describing isn't really part of ZEO, it's really more in the realm of what is already in Zope. Thus, if you're eager to add such a feature, you already have what you need. Start tinkering with the transaction machinery and see how you can extend it. --Paul
-----Original Message----- From: Andrew Kenneth Milton [mailto:akm@mail.theinternet.com.au] Sent: Wednesday, March 22, 2000 8:06 PM To: Michel Pelletier Cc: Andrew Kenneth Milton; Paul Everitt; zope@zope.org Subject: Re: [Zope] ANN: ZEO goes Open Source
+----[ Michel Pelletier ]--------------------------------------------- | | > Is the ZSS replication and fail-over machinery in place yet? | | No. These are very deep issues, especially the replication part. If | this is important to you, I suggest using a replicated filesystem to | store the final FileStorage file, like Coda or Intermetzo.
Some sort of transaction log would be nice. Not only could you track this to do replication, it would then possible to rebuild corrupted Data.fs files as well. I don't know if ZEO already does this, this is just idle speculation on my part.
| While it would be a great idea for there to be failover and replication, | the issues are complex and without more wide spread use we really don't | want to start engineering a solution (one of the reasons we are open | sourcing it...)
I guessed as much :-)
-- Totally Holistic Enterprises Internet| P:+61 7 3870 0066 | Andrew Milton The Internet (Aust) Pty Ltd | F:+61 7 3870 4477 | ACN: 082 081 472 | M:+61 416 022 411 | Carpe Daemon PO Box 837 Indooroopilly QLD 4068 |akm@theinternet.com.au|
+----[ Paul Everitt ]--------------------------------------------- | | Note that what you're describing isn't really part of ZEO, it's really | more in the realm of what is already in Zope. Thus, if you're eager to | add such a feature, you already have what you need. Start tinkering | with the transaction machinery and see how you can extend it. Except at the ZEO level, the conflicts must have been resolved. I only want to replicate the state of play at the ZSS end. Otherwise I could alter ZEO to take a list of ZSSs and do a 'broadcast' and have them all do the same work (hoping like crazy that all transactions arrived at all servers in the same order). Then I would have n copies of the primary ZSS. Failover from this point is almost trivial. The other option is to get the ZSS to forward transactions in the order IT gets them and send them to a list of servers it knows about. This is slightly cleaner, since all of the Zope machines don't need to know about all of the ZSSs. The other major problem is, that if I do this work, I will be unable to contribute it back to the project for some time (if ever). I don't really have the facilities at home to be able to do ZEO type work, my current employers (where I have more available resources) like OSS, but, tend to treat it as a code repository, rather than a symbiotic relationship. -- Totally Holistic Enterprises Internet| P:+61 7 3870 0066 | Andrew Milton The Internet (Aust) Pty Ltd | F:+61 7 3870 4477 | ACN: 082 081 472 | M:+61 416 022 411 | Carpe Daemon PO Box 837 Indooroopilly QLD 4068 |akm@theinternet.com.au|
participants (2)
-
Andrew Kenneth Milton -
Paul Everitt