| 1 |
Am Mittwoch 03 Juni 2009 schrieb Florian Philipp: |
| 2 |
> |
| 3 |
> Do you have a spare network adapter, maybe an older 100MBit PCI card? |
| 4 |
> Maybe we should rule out a hardware fault on your ethernet chipset first. |
| 5 |
> |
| 6 |
I already thought on this, but the results of my tests dont indicate a |
| 7 |
hardware fault on the ethernet chipset, because: |
| 8 |
|
| 9 |
* I can run a ping -f to the machine, it runs for hours without the |
| 10 |
slightest problem |
| 11 |
* As long as files transfered are small enough (i.e. they fit in the cache |
| 12 |
buffer on the server) and the server has enough time to write back it to |
| 13 |
the disk, there is no problem |
| 14 |
* If I explicitly force the ethernet link to be 100FD instead of gigabit, |
| 15 |
the is also no problem. So I don't expect any error using another 100MBit |
| 16 |
card. |
| 17 |
|
| 18 |
For me it looks like as if the following is happening: |
| 19 |
|
| 20 |
* Memory gets filled up with cached files, no problem so far |
| 21 |
* If no more physical ram is available, the system tries to free some memory |
| 22 |
internally, e.g. by flushing the caches. |
| 23 |
* If releasing cache entries and writing back data to their respective |
| 24 |
files does not perform fast enough, an internal memory allocation may not |
| 25 |
succeed, and I see the "page allocation failure" messages, with different |
| 26 |
processes/kernel threads in the first line. |
| 27 |
* I assume that most of the internal kernel threads don't get a problem in |
| 28 |
this situation, but there may be some critical parts where we do. Hence, it |
| 29 |
might just be a matter of probability whether it encounters such a critical |
| 30 |
part, and the probabilty increases with the MB/s the data is put to the NFS |
| 31 |
server. |
| 32 |
|
| 33 |
Greetings |
| 34 |
Alex |
Archives