Hi Ajanthan,
On Wed, Feb 12, 2014 at 2:59 PM, bala ajanthan <[email protected]>wrote: > > > > On Wed, Feb 12, 2014 at 11:02 AM, Isuru Haththotuwa <[email protected]>wrote: > >> Hi Devs, >> >> The purpose of this thread is to do discuss $subject. >> >> To monitor the status and debug in case of an issue, its very helpful if >> some information about the instance can be collected. Specially if the >> instance crashed due to an issue, there will be no information available on >> what went wrong within the instance. >> >> As the initial version, I propose to monitor several logs in the instance >> and publish them to an external monitoring server where the logs will be >> collected. I have attached the proposed sequence diagram here: >> >> [image: Inline image 1] >> >> Here, a user would specify the log file locations when a cartridge is >> deployed (as properties?). This information would be extracted by the >> Stratos Manager and put to the payload that is sent to the instance via >> Cloud Controller. The Cartridge Agent in the instance would read the >> relevant log files and send them to a monitoring server. Please share your >> thoughts. >> > Reading files time to time and sending to external sinks seems to be not > good idea.Log collecting should be completely event driven which will > enable us to give real time logs to user. > How about Agent acting as syslog server?Makes the agent to receive syslog > events within the instance and then sending those events to pluggable > distributed logging solutions.Apache fume[0] seems to be good fit. > It can act as a bridge between log sources and sinks.It has syslog > sources[1] for collecting syslog events and various sinks. > Thanks for the input. I also considered the use of syslog for this. But there are issues in configuring forwarding logs to syslog. For example, I was unable to configure mysql to send all required logs to syslog. Another issue is that we have to do custom configurations for different cartridges types to send logs to syslog. Still, I too think syslog would be an ideal solution for this if we can get past these issues. > >> -- >> Thanks and Regards, >> >> Isuru H. >> +94 716 358 048* <http://wso2.com/>* >> >> >> > [0]http://flume.apache.org/FlumeUserGuide.html#data-flow-model > [1]http://flume.apache.org/FlumeUserGuide.html#syslog-sources > > -- > Balachandiran Ajanthan > Phone:+94775581497 > Blog : http://ajanthan87.blogspot.com > Twitter: ajanthan87 > skype :bala.ajanthan > gtalk :balaajanthan > -- Thanks and Regards, Isuru H. +94 716 358 048* <http://wso2.com/>*
<<inline: log_publisher_sequence.png>>
