So.. We´ve setup a montior for a customer that check the content of a file to determine of the server should be Online / Offline. This file is edited manually so that they themselves can control the state of the pool member.
I´ve tested both a standard https and a reverse, the reverse seems to fit the task.. except one thing, if the server that hosts the status files goes offline.. all servers in the pool go offline (and thats is a problem for both standard and reverse)..
Anyone got any ides / tips / suggestions on how to accomplish something like this?
We dont have a bigiq in place yet, we would like to avoid partitions as well... so something done without access to bigip/bigiq.
Hi @kimhenriksen ,
> If I understand you well , using Reverse option in your environment should not be fit , as reverse marks the pool member down if it matches with "the correct Receive string/content" , Reverse option is very useful for instance if you want to match with " ERROR 404 " so when the web page reurned " Error 404 " monitor marks pool member down.
> so explain , How do you configure Reverse monitor ?
Also clarify what is needed from F5 to match your needs ?
Ok, so the way i´ve configured it in my lab is like this.
I have a pool of 3 server, two monitors. One monitor checks the health of each member using a http header and that works very well. The second (the reverse monitor) checks that a status file (with an alias on the monitor to check it on another server, sort of a status mgmt server for the pool members) does not contain the string Maintenance, if it does it should be considered offline (not disabled) and a new member will be reselected.
But.. I wondered if it is possible to have it react only of the recieve string and sort of disable the timeout setting completly. This way i could hand over control of the pool to a party that doesnt have access to the ADC platform without the risk of losing the whole pool if the status mgmt server went offline.
For this inquiry : " But.. I wondered if it is possible to have it react only of the recieve string and sort of disable the timeout setting completly. This way i could hand over control of the pool to a party that doesnt have access to the ADC platform without the risk of losing the whole pool if the status mgmt server went offline. " )
Do you mean that you do not want to loose the avilabilty of pool if the " Reverse" monitor fails/down.
you mean that , it is of course doable.
- Another thing :
I understood from you that you want to prevent these users from modifying in " Content of file " that you are assigning your monitor.
- Or you can replace the two monitors by using this
Configure ( send string , Receive string and Recieve disable string as well )
"Recieve disable string" means if this string returned from pool member , this monitor should be down.
I hope I could help.
Ok, any tips on how the timeout disable is doable? I´ve yet to find a solution.
Yes, if we comebine those two to one.. we´ll might get a problem with the Get URI.. they might have to be independent of each other. And the filecheck would be good if it could be done to another IP then the pool member. The customer want the member as standard as possible, but still have the possibility for online/offline.
But.. If the first part is solved, that solves all of it.
Yeah, but if configured that way.. you loose the ability of monitor1 to function correctly. Monitor1 should always take precedence in my case as that checks the actually health of the server. If i could specify that per monitor that would be great.
if you want to make Monitor 1 to take an effect and not forward traffic to its pool_members , you will need to assign the two monitor and Change Availability Requirement from at "least" to "ALL" or " at "Least = 2 ".
> What about trying monitors per node it self :
> you do not reply about , what do you mean by disabling timeouts.
yes , If you configure " Availability Requirement " to "ALL" and one of your assigned monitors fails " the whole node will be marked down >>> this is for Nodes level.
- If you did this on Pool Level , and " Availability Requirement " to "ALL" and one of your assigned monitors fails " the whole Pool will be marked down and wouldn’t recieve anymore traffic and this will appeare that " the impacted virtual server will reset the traffic that destinated to with a reason of ( no pool_members available ) in reset packets "
yes , so
could you please figure out your design/or simplify you needs to search for a workaround for your environment.
because I miss much points from your design , it is not very clear to me.
okay , you have one pool contains , two members.
> Do you assign both of monitors on pool Level , or on Nodes "members" level ?
> Do you loadbalance traffic to both of them ? , also maybe what makes the monitor down " All servers Marked down " the server owner or what ever the person who change the "File" that you monitor on tampers this file on both of servers so you see both of them marked down not only one of them .
> your Request is :
you want to detect the change of this file , or a way to restrict this file on server and not make anyone to tamper it.
Specifiy your request more
I’m looking forward to your reply.
you can not disable the Timeout.
you can put a very large value instead of "16" seconds (default) , you can put "86400" this value.
that is mean your monitor will not be marked down for a full day "86400" seconds , it means almost you disabled this timeout.
I do not why you want to do that really ?