2004-04-05, h keltezéssel 01:53-kor Jason Haar ezt írta:
Hi there
We're running a multi-country network of syslog-ng with centralized logging (via TCP).
However, due to the possibility of network failure, I am not willing to rely on the central server for triggering alerts - so we do that on the actual leaf syslog servers instead. However, it would be much better if we could actually centralize our alerting system (due to what we are alerting about).
So I was wondering if syslog-ng's TCP transport layer could be turned into something like MySQL's master-slave mode: write to a cache file first, then write to remote server, and if that succeeds, remove from cache file.
That way if a network outage occurred, the cache file would just get larger and larger, then when the network came up again, it would flow through without any records being lost.
[I know the fifo option allows this in RAM - I guess all that I'm wondering about is could that be done on disk instead?. A cache file would survive restarts/etc.]
sorry, this is not currently possible. -- Bazsi PGP info: KeyID 9AF8D0A9 Fingerprint CD27 CFB0 802C 0944 9CFD 804E C82C 8EB1