responses
I have the usual error "Unexpected data from server ..." using om_http to send Windows event logs through Nginx to Redis.
The problem is that when randomly this error happens ( once every 5 days) the service unexpectently stops !
From the source code I can understand that NXLog receives a response which cannot map to a correnponding request.
It has a response with an appropriate length but the corresponding request is null! Possibly is a bad logic implementation, or even a network problem that could cause request/response mismatch.I really don't know.
But I also really can't understand why NxLog preferres to die instead of just return from the function in case this happens and not execute the code following the condition that drives the service to crash at the end.
Stelios
Comments (5)
Thanks for your reply.
I will ask and deploy the EE trial and inform if the fix solves this.
I have got and installed NXLog EE trial.
In the Changelog file confirms that it has applied the fix:
Fixed apr_pollset_poll and "unexpected data from server" errors in om_http when the remote server sent fragmented data.
But om_http.dll is missing from the deployed modules after the installation.
So I cannot actually test it.
Is the missing module one of the limitations of the trial version or an error of the setup packaging, and what are the trial limitations ?
Kind Regards
Stelios
The om_http.dll module should be there. Please try to remove nxlog-ce before installing the EE, otherwise the repair option should also help.
There are no limitations in the trial.
Simple things in life are the most difficult.
om_http now connects immediately to the host specified (and not on event arrived) and up to now there is no unexpected response message that crashes nxlog.
I will test it some days anyway, because I cannot understand how and when the server sends fragmented data.
After several days of testing I confirm that the fix applied on the om_http module in the nxlog-2.10.1662-trial solves the "unexpected data received from serrver" error and the service crash caused from random generated fragmented server responses.