No relevant resource is found in the selected language.

This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>Search


To have a better experience, please upgrade your IE browser.


Eudemon200 rearrange two-node cluster hot backup frequently lead the firewall cannot save configuration

Publication Date:  2012-09-25 Views:  2 Downloads:  0

Issue Description

Some office abroad feed back the information which the firewall cannot save configuration ,there is much more space when we check the remain memory and the memory cannot be divulged.After asking the local employee their early operation ,we get an important information.One month ago,the network is error and create ring network,it lead firewall rearranges two-node cluster hot backup frequently.The problem has been solved by adjusting the network.We find it cannot save configuration when we configure the firewall after period of time.
Alarm information:
Check the information statistics of hrp:
HRP_S[smsc&mmsc-E200(1)-hidecmd]dis hrp st                                     
HRP current queue length = 0                                                  
HRP queue full discard   = 583   ----this queue is full   
HRP baksession AgedCount = 14                                          
HRP bakSevMap  AgedCount = 0                                     
HRP Sock task send times = 829

Alarm Information


Handling Process

This problem has been led by memory separation ,it can be solved by resetting firewall.
After restarting,we should configure firewall again and save the configuration.

Root Cause

When the firewall saves configuration ,it will apply for a piece of memory to cache the configuration.When it fail to apply for memory,it cannot save configuration.We can find the firewall memory has been parted many pieces which has 2048 byte by analysing the firewall collection information.There is no big memory to supply to firewall,it lead when we save the configuration ,the memory application is failed.
The raeson that the memory is parted to small piece is the memory has been used by small piece.Based on checking the history memory,we can find the hrp tasks apply for all the memory,the size of hrp information is 2048 and the information queue of hrp has been occupied completely.Thereout we can know ,the firewall exhaust the memory because of two-node cluster hot backup and it is used to make the batch backup of two-node cluster hot backup( the memory content is configuration data).Because the firewall is composed ring network based on the frontline feelback ,thereout we can know because of the two-node cluster hot backup network ,it leads primary and secondary oscillation and primary and secondary rearrange frequently.When the primary and secondary are rearranging,the primary host will make the configuration backup to secondary one.Before the firewall finishing the backup information , the primary and secondary rearrange again and again ,it will lead the hrp information of firewall  accumulates more and more and exhausts all the memory.
When the problem has been solved and firewall finishes doing with the hrp information ,the memory has been parted many pieces of 2048.After the memory has been released,if it satisfies the reassembly condition,the small pieces memory will combine to a piece of big memory.
Memory conformity is based on the continuous memory which has been released to progress assigning,some continuous memory which has been released will combine to a piece of big memory again.If there is no fixed memory which has been released in the memory(such as dispensing configuration seizes memory and so on),it will lead the memory cannot be continuous and cannot combine to big memory again.In total the situation which there are too many memory fragments to conformity is very few .So it will appear the situation which the primary firewall applys for memory but cannot save configuration, the memory which has been released by secondary firewall can conformity again ,it can save configuration normally.