| 1 |
Steve, |
| 2 |
|
| 3 |
>> |
| 4 |
>> so unless it's of real use to others I might just continue to support it myself. |
| 5 |
>> I'm already packaging up other internal apps using a portage overlay. |
| 6 |
>> |
| 7 |
|
| 8 |
I had several clients deploying mysql server clusters in large webfarms at the |
| 9 |
start of this year and investigated the options for commercial support of mysql |
| 10 |
under gentoo at that time. This investigation included attending the mysql |
| 11 |
2004 user conference and talking to several decision makers within mysql |
| 12 |
along with conversations with other gentoo developers and some discussion |
| 13 |
on the gentoo-dev ml. |
| 14 |
|
| 15 |
To summarize: |
| 16 |
a) Mysql would not support gentoo unless the tree was somehow stable. This |
| 17 |
meant either GLEP19 would have to become a reality or I would need to maintain |
| 18 |
a generic commercial mysql profile/overlay for the indefinite future. If I could |
| 19 |
provide either of these options, mysql would perform their own testing and eventually |
| 20 |
add gentoo to the list of supported platforms. |
| 21 |
b) Waiting or assisting GLEP19 seemed to be unlikely to reach maturity in any reasonable |
| 22 |
time so I dropped that approach. |
| 23 |
c) I started setting up the overlay/profile and commited some initial revisions to cvs, but |
| 24 |
after talks with wolf10k2 and the rest of the releng team was convinced that I was getting |
| 25 |
myself into more work than I could handle on my own. |
| 26 |
|
| 27 |
At the same time that I was working on the above, mysql was announcing partnerships with |
| 28 |
Dell and Redhat, such that if a commercially supported server running those platforms ever |
| 29 |
went down, mysql would take over complete responsibility for bringing the system back |
| 30 |
online (even if that meant bringing in dell and redhat techs to form a combined force). |
| 31 |
Eventually my clients decided that the cost/benefit ratio of maintaining a 100% gentoo |
| 32 |
server farm vs a hybrid redhat on db servers, gentoo on everything else farm was such |
| 33 |
that the hybrid approach was the only realistic course to take. |
| 34 |
|
| 35 |
As a systems administrator and developer, the result was a real frustration for me - but |
| 36 |
sometimes you have to live with such compromises. |
| 37 |
|
| 38 |
I thought I had removed the commercial-mysql profile from CVS. If I havent, I sincerely |
| 39 |
apologize and will clean it up asap. |
| 40 |
|
| 41 |
>> FYI - if you deploy an app which is non-gpl which links against mysql libs then |
| 42 |
>> you need to deploy against a commercial version of mysql. |
| 43 |
>> this is built and maintained in binary format by mysql - and requires |
| 44 |
>> username/password to download. |
| 45 |
>> |
| 46 |
|
| 47 |
OK. It sounds like you really dont care about their supporting the binaries, you are |
| 48 |
just trying to fulfill your legal responsibilities. In that case, perhaps it does make |
| 49 |
sense to add some kind of new ebuild or profile. |
| 50 |
|
| 51 |
However, I dont think it would have broad appeal. The source built standard gentoo |
| 52 |
mysql ebuilds are quite good. In fact, I prefer them over having to maintain the commercial |
| 53 |
ones. Probably 99% of the interest in commercial binaries is for support, especially with |
| 54 |
mysql 5 and database clusters. I dont see your ebuilds or profiles really addressing that. |
| 55 |
|
| 56 |
Sincerely, |
| 57 |
Matt |
| 58 |
-- |
| 59 |
gentoo-dev@g.o mailing list |
Archives