1 |
Eric Brown wrote: |
2 |
> Most people can probably get away with using chroots instead of |
3 |
> completely separate machines for testing. The only major concern |
4 |
> there is whether or not you have the computing resources available to |
5 |
> build on the same host that your production applications run on. |
6 |
|
7 |
I interpreted corporate to mean "everyone is gone by 6pm" so it's roll |
8 |
in around 11am on Tuesday and do an update starting around 6:30pm that |
9 |
night. However that might not be the case. I've never had issues doing |
10 |
updates in off hours usually 1-3am on lightly loaded servers, but that |
11 |
depends on your hardware and workload. |
12 |
|
13 |
Here are some compile times since I was playing with some new boxes today. |
14 |
|
15 |
3.0Ghz Xeon w/2MB cache w/HT, 1GB RAM, single drive |
16 |
time to compile Mysql 5.0.19 w/-j3 - 23 minutes |
17 |
|
18 |
2x 3.0Ghz Xeon w/2MB cache w/HT, 4GB RAM, raid 1 w/256 cache |
19 |
time to compile Mysql 5.0.19 w/-j3 - 9 minutes |
20 |
|
21 |
2x dual core 2.8Ghz Xeon w/2MB cache/core w/HT, 8GB RAM, raid1 w/256 cache |
22 |
time to compile Mysql 5.0.19 w/-j8 - 7 minutes |
23 |
|
24 |
In the last case I could renice or use -j5 and load would stay under 3 |
25 |
meaning the OS could have the remaining resources which would likely be |
26 |
enough in off hours. However some of these boxes may be of a slightly |
27 |
different scale than other admins |
28 |
|
29 |
I am curious about how other people do updates once they've decided |
30 |
there is no issue and how hardware/load plays a part in the process. |
31 |
|
32 |
kashani |
33 |
-- |
34 |
gentoo-server@g.o mailing list |