Re: Content Switching and Keepalives [7:43141]
Thanks for the info everyone, I tested it last night and it worked great, we now have load balancing and the keepalive running. Here's the config for one of the services Cheers Pat service portal2 ip address 172.16.10.12 string portal2 protocol tcp keepalive port 81 keepalive type http keepalive uri "/index.html" active sam sneed wrote: > There are 2 methods of keepalives, get and head. > > get: > CSS gets the web page, computes a hash based on the page and stores it for > reference. The next time the CSS gets the webpage it looks for 200 OK and > stauts and compares the new hash with the hash stored for reference. If they > are different the CSS marks the service as down. So you can conclude this > method only works well for static content on pages. > > Head: > CSS only issues an HTTP head on the service and looks for 200 OK status , if > it gets it service is marked up other wise its down. Less overhead than get > method and good for Dynamic content as well. > > hope that helped a bit. > > ""Patrick Donlon"" wrote in message > [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > > Hi Dave > > > > I've not had chance to test the keepalive yet but I see you mention using > > head or get > > can depend on the page type. Can you explain further or do you have any > > links? > > > > Cheers > > > > Pat > > > > David Harrison wrote: > > > > > This is correct. The domain name is not necessary. Since the CSS knows > > > the ip address of the box it's watching it doesn't have to rely on a > > > domain name to find the location of the server. > > > > > > However it is important that the css know the path to reach the > > > reference page. > > > > > > I've used the following: > > > service blah_blah > > > ip address 10.1.1.1 > > > keepalive frequency 8 > > > keepalive type http > > > keepalive uri "/.reference/arrowpoint-keepalive.html" > > > active > > > > > > I usually use the default "head" method vs the "get". Depends on whether > > > the file you are watching is static or dynamic. > > > > > > Dave > > > > > > -Original Message- > > > From: John Neiberger [mailto:[EMAIL PROTECTED]] > > > Sent: Friday, May 03, 2002 12:19 PM > > > To: [EMAIL PROTECTED] > > > Subject: Re: Content Switching and Keepalives [7:43141] > > > > > > I'm not positive about this but I don't believe you're supposed to > > > include the domain name in the URI. We simply use 'keepalive uri > > > "/index.htm"' and that works well. Give that a shot and see if it works > > > for you. > > > > > > John > > > > > > >>> "Patrick Donlon" 5/3/02 9:54:47 AM >>> > > > Hi > > > > > > I tested it and for some reason it didn't work, I configured the > > > following > > > on the > > > service: > > > > > > keepalive port 81, > > > keepalive method get, > > > keepalive type http > > > keepalive frequency 25, > > > keepalive retry 25 > > > keepalive uri "www.blahblah.com/index.html" > > > > > > I then activated the service (and re-activated it a few times just in > > > case) > > > Any thing > > > obviously wrong and what should I check in the log > > > > > > cheers > > > > > > Pat > > > > > > Patrick Donlon wrote: > > > > > > > Hi All > > > > > > > > I have two web servers which are being load balanced behind a CSS, > > > this > > > > is working fine. Currently we're using the default ICMP keepalive, > > > this > > > > is OK if the failure is at this level but when the web services > > > process > > > > is stopped by the DBA the CSS thinks it's up and running. I've seen > > > the > > > > different options, tcp, http gets, etc, and would like to know > > > anyone > > > > else's experience in what is the best balance over performance and > > > > detecting the lost of service > > > > > > > > Cheers > > > > > > > > Pat > > > > > > > > [EMAIL PROTECTED] Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=43475&t=43141 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
Re: Content Switching and Keepalives [7:43141]
There are 2 methods of keepalives, get and head. get: CSS gets the web page, computes a hash based on the page and stores it for reference. The next time the CSS gets the webpage it looks for 200 OK and stauts and compares the new hash with the hash stored for reference. If they are different the CSS marks the service as down. So you can conclude this method only works well for static content on pages. Head: CSS only issues an HTTP head on the service and looks for 200 OK status , if it gets it service is marked up other wise its down. Less overhead than get method and good for Dynamic content as well. hope that helped a bit. ""Patrick Donlon"" wrote in message [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > Hi Dave > > I've not had chance to test the keepalive yet but I see you mention using > head or get > can depend on the page type. Can you explain further or do you have any > links? > > Cheers > > Pat > > David Harrison wrote: > > > This is correct. The domain name is not necessary. Since the CSS knows > > the ip address of the box it's watching it doesn't have to rely on a > > domain name to find the location of the server. > > > > However it is important that the css know the path to reach the > > reference page. > > > > I've used the following: > > service blah_blah > > ip address 10.1.1.1 > > keepalive frequency 8 > > keepalive type http > > keepalive uri "/.reference/arrowpoint-keepalive.html" > > active > > > > I usually use the default "head" method vs the "get". Depends on whether > > the file you are watching is static or dynamic. > > > > Dave > > > > -Original Message- > > From: John Neiberger [mailto:[EMAIL PROTECTED]] > > Sent: Friday, May 03, 2002 12:19 PM > > To: [EMAIL PROTECTED] > > Subject: Re: Content Switching and Keepalives [7:43141] > > > > I'm not positive about this but I don't believe you're supposed to > > include the domain name in the URI. We simply use 'keepalive uri > > "/index.htm"' and that works well. Give that a shot and see if it works > > for you. > > > > John > > > > >>> "Patrick Donlon" 5/3/02 9:54:47 AM >>> > > Hi > > > > I tested it and for some reason it didn't work, I configured the > > following > > on the > > service: > > > > keepalive port 81, > > keepalive method get, > > keepalive type http > > keepalive frequency 25, > > keepalive retry 25 > > keepalive uri "www.blahblah.com/index.html" > > > > I then activated the service (and re-activated it a few times just in > > case) > > Any thing > > obviously wrong and what should I check in the log > > > > cheers > > > > Pat > > > > Patrick Donlon wrote: > > > > > Hi All > > > > > > I have two web servers which are being load balanced behind a CSS, > > this > > > is working fine. Currently we're using the default ICMP keepalive, > > this > > > is OK if the failure is at this level but when the web services > > process > > > is stopped by the DBA the CSS thinks it's up and running. I've seen > > the > > > different options, tcp, http gets, etc, and would like to know > > anyone > > > else's experience in what is the best balance over performance and > > > detecting the lost of service > > > > > > Cheers > > > > > > Pat > > > > > > [EMAIL PROTECTED] Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=43420&t=43141 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
Re: Content Switching and Keepalives [7:43141]
Hi Dave I've not had chance to test the keepalive yet but I see you mention using head or get can depend on the page type. Can you explain further or do you have any links? Cheers Pat David Harrison wrote: > This is correct. The domain name is not necessary. Since the CSS knows > the ip address of the box it's watching it doesn't have to rely on a > domain name to find the location of the server. > > However it is important that the css know the path to reach the > reference page. > > I've used the following: > service blah_blah > ip address 10.1.1.1 > keepalive frequency 8 > keepalive type http > keepalive uri "/.reference/arrowpoint-keepalive.html" > active > > I usually use the default "head" method vs the "get". Depends on whether > the file you are watching is static or dynamic. > > Dave > > -Original Message- > From: John Neiberger [mailto:[EMAIL PROTECTED]] > Sent: Friday, May 03, 2002 12:19 PM > To: [EMAIL PROTECTED] > Subject: Re: Content Switching and Keepalives [7:43141] > > I'm not positive about this but I don't believe you're supposed to > include the domain name in the URI. We simply use 'keepalive uri > "/index.htm"' and that works well. Give that a shot and see if it works > for you. > > John > > >>> "Patrick Donlon" 5/3/02 9:54:47 AM >>> > Hi > > I tested it and for some reason it didn't work, I configured the > following > on the > service: > > keepalive port 81, > keepalive method get, > keepalive type http > keepalive frequency 25, > keepalive retry 25 > keepalive uri "www.blahblah.com/index.html" > > I then activated the service (and re-activated it a few times just in > case) > Any thing > obviously wrong and what should I check in the log > > cheers > > Pat > > Patrick Donlon wrote: > > > Hi All > > > > I have two web servers which are being load balanced behind a CSS, > this > > is working fine. Currently we're using the default ICMP keepalive, > this > > is OK if the failure is at this level but when the web services > process > > is stopped by the DBA the CSS thinks it's up and running. I've seen > the > > different options, tcp, http gets, etc, and would like to know > anyone > > else's experience in what is the best balance over performance and > > detecting the lost of service > > > > Cheers > > > > Pat > > > > [EMAIL PROTECTED] Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=43380&t=43141 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
Re: Content Switching and Keepalives [7:43141]
A little off topic, but has anybody tried similar with Foundry. Got it to work yet? Got it to fail properly when the service is unavailable or does it pass the ICMP health check, then fail the layer 7 check and flap forever. Can't believe how flaky Foundry is turning out to be recently. I've got a bee in my bonnet... Layer 3 filtering on Foundry NetIrons contained in all the documentation. Tried to implement it, didn't work, reported to Foundry. Informed that NetIrons do not now support it. Sure enough, the documentation was almost immediately updated. They've just binned it - one of many. Bit late when you've sold it. Tried adding/removing ports from VLAN's with the web management. Doh! We're looking at stripping out ServerIrons and putting in CSS, partly due to the extra flexibility available with the Cisco health checking, but mainly because the support works! You can knock Cisco all you like, but we keep coming back to them. Anybody got any views on this? Gaz ""David Harrison"" wrote in message [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > This is correct. The domain name is not necessary. Since the CSS knows > the ip address of the box it's watching it doesn't have to rely on a > domain name to find the location of the server. > > However it is important that the css know the path to reach the > reference page. > > I've used the following: > service blah_blah > ip address 10.1.1.1 > keepalive frequency 8 > keepalive type http > keepalive uri "/.reference/arrowpoint-keepalive.html" > active > > I usually use the default "head" method vs the "get". Depends on whether > the file you are watching is static or dynamic. > > Dave > > > -----Original Message- > From: John Neiberger [mailto:[EMAIL PROTECTED]] > Sent: Friday, May 03, 2002 12:19 PM > To: [EMAIL PROTECTED] > Subject: Re: Content Switching and Keepalives [7:43141] > > I'm not positive about this but I don't believe you're supposed to > include the domain name in the URI. We simply use 'keepalive uri > "/index.htm"' and that works well. Give that a shot and see if it works > for you. > > John > > >>> "Patrick Donlon" 5/3/02 9:54:47 AM >>> > Hi > > I tested it and for some reason it didn't work, I configured the > following > on the > service: > > keepalive port 81, > keepalive method get, > keepalive type http > keepalive frequency 25, > keepalive retry 25 > keepalive uri "www.blahblah.com/index.html" > > I then activated the service (and re-activated it a few times just in > case) > Any thing > obviously wrong and what should I check in the log > > cheers > > Pat > > > > > Patrick Donlon wrote: > > > Hi All > > > > I have two web servers which are being load balanced behind a CSS, > this > > is working fine. Currently we're using the default ICMP keepalive, > this > > is OK if the failure is at this level but when the web services > process > > is stopped by the DBA the CSS thinks it's up and running. I've seen > the > > different options, tcp, http gets, etc, and would like to know > anyone > > else's experience in what is the best balance over performance and > > detecting the lost of service > > > > Cheers > > > > Pat > > > > [EMAIL PROTECTED] Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=43272&t=43141 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
RE: Content Switching and Keepalives [7:43141]
This is correct. The domain name is not necessary. Since the CSS knows the ip address of the box it's watching it doesn't have to rely on a domain name to find the location of the server. However it is important that the css know the path to reach the reference page. I've used the following: service blah_blah ip address 10.1.1.1 keepalive frequency 8 keepalive type http keepalive uri "/.reference/arrowpoint-keepalive.html" active I usually use the default "head" method vs the "get". Depends on whether the file you are watching is static or dynamic. Dave -Original Message- From: John Neiberger [mailto:[EMAIL PROTECTED]] Sent: Friday, May 03, 2002 12:19 PM To: [EMAIL PROTECTED] Subject: Re: Content Switching and Keepalives [7:43141] I'm not positive about this but I don't believe you're supposed to include the domain name in the URI. We simply use 'keepalive uri "/index.htm"' and that works well. Give that a shot and see if it works for you. John >>> "Patrick Donlon" 5/3/02 9:54:47 AM >>> Hi I tested it and for some reason it didn't work, I configured the following on the service: keepalive port 81, keepalive method get, keepalive type http keepalive frequency 25, keepalive retry 25 keepalive uri "www.blahblah.com/index.html" I then activated the service (and re-activated it a few times just in case) Any thing obviously wrong and what should I check in the log cheers Pat Patrick Donlon wrote: > Hi All > > I have two web servers which are being load balanced behind a CSS, this > is working fine. Currently we're using the default ICMP keepalive, this > is OK if the failure is at this level but when the web services process > is stopped by the DBA the CSS thinks it's up and running. I've seen the > different options, tcp, http gets, etc, and would like to know anyone > else's experience in what is the best balance over performance and > detecting the lost of service > > Cheers > > Pat > > [EMAIL PROTECTED] Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=43240&t=43141 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
Re: Content Switching and Keepalives [7:43141]
I'm not positive about this but I don't believe you're supposed to include the domain name in the URI. We simply use 'keepalive uri "/index.htm"' and that works well. Give that a shot and see if it works for you. John >>> "Patrick Donlon" 5/3/02 9:54:47 AM >>> Hi I tested it and for some reason it didn't work, I configured the following on the service: keepalive port 81, keepalive method get, keepalive type http keepalive frequency 25, keepalive retry 25 keepalive uri "www.blahblah.com/index.html" I then activated the service (and re-activated it a few times just in case) Any thing obviously wrong and what should I check in the log cheers Pat Patrick Donlon wrote: > Hi All > > I have two web servers which are being load balanced behind a CSS, this > is working fine. Currently we're using the default ICMP keepalive, this > is OK if the failure is at this level but when the web services process > is stopped by the DBA the CSS thinks it's up and running. I've seen the > different options, tcp, http gets, etc, and would like to know anyone > else's experience in what is the best balance over performance and > detecting the lost of service > > Cheers > > Pat > > [EMAIL PROTECTED] Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=43233&t=43141 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
Re: Content Switching and Keepalives [7:43141]
Hi I tested it and for some reason it didn't work, I configured the following on the service: keepalive port 81, keepalive method get, keepalive type http keepalive frequency 25, keepalive retry 25 keepalive uri "www.blahblah.com/index.html" I then activated the service (and re-activated it a few times just in case) Any thing obviously wrong and what should I check in the log cheers Pat Patrick Donlon wrote: > Hi All > > I have two web servers which are being load balanced behind a CSS, this > is working fine. Currently we're using the default ICMP keepalive, this > is OK if the failure is at this level but when the web services process > is stopped by the DBA the CSS thinks it's up and running. I've seen the > different options, tcp, http gets, etc, and would like to know anyone > else's experience in what is the best balance over performance and > detecting the lost of service > > Cheers > > Pat > > [EMAIL PROTECTED] Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=43232&t=43141 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
Re: Content Switching and Keepalives [7:43141]
Hi, You should change the keepalive type in the content switch to HTTP or SSL which ever your webserver is using.For example it should be like the one below. For http service webserver1 ip address x.x.x.x keepalive port 80 keepalive type http active For SSL ervice webserver1 ip address x.x.x.x keepalive port 443 keepalive type tcp active Kind Regards /Thangavel 186K Reading,Brkshire Direct No -0118 9064259 Mobile No -07796292416 Post code: RG16LH www.186k.co.uk -- The greatest glory in living lies not in never falling, but in rising every time we fall ." -- Nelson Mandela "Patrick Donlon" To: [EMAIL PROTECTED] Fax to: Sent by: Subject: Content Switching and Keepalives [7:43141] nobody@groupst udy.com 02/05/2002 18:15 Please respond to "Patrick Donlon" Hi All I have two web servers which are being load balanced behind a CSS, this is working fine. Currently we're using the default ICMP keepalive, this is OK if the failure is at this level but when the web services process is stopped by the DBA the CSS thinks it's up and running. I've seen the different options, tcp, http gets, etc, and would like to know anyone else's experience in what is the best balance over performance and detecting the lost of service Cheers Pat [EMAIL PROTECTED] ** This e-mail is from 186k Ltd and is intended only for the addressee named above. As this e-mail may contain confidential or priveleged information, if you are not the named addressee or the person responsible for delivering the message to the named addressee, please advise the sender by return e-mail. The contents should not be disclosed to any other person nor copies taken. 186k Ltd is a Lattice Group company, registered in England & Wales No. 3751494 Registered Office 130 Jermyn Street London SW1Y 4UR ** Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=43214&t=43141 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
Re: Content Switching and Keepalives [7:43141]
I have a pair of CS11152 in HA mode and we use HTTP for load balancing. In my opinion if your are using ping to load balance it is a waste. The CSS's are very capable and load balance HTTP very well. As far as load balancing methods the engineer at TAC said to use round robin which is the deafult scheme. I was using ACA, but the engineer said that should be only used for the heaviest fo traffic sites so i recommend using round robin as well, it'lll work better. Here is a sample config: service svc-w1.web ip address 172.16.10.100 port 80 keepalive type http keepalive method get keepalive uri "/http-ping.html" active service svc-w2.web ip address 172.16.10.101 port 80 keepalive type http keepalive method get keepalive uri "/http-ping.html" active content cnt-www.cobrand protocol tcp port 80 balance roundrobin ##you really don;t need this line becuase roundrobin is defaut. url "/*" add service svc-w1.web add service svc-w2.web vip address 192.168.133.100 active hope this helps ""Patrick Donlon"" wrote in message [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > Hi All > > I have two web servers which are being load balanced behind a CSS, this > is working fine. Currently we're using the default ICMP keepalive, this > is OK if the failure is at this level but when the web services process > is stopped by the DBA the CSS thinks it's up and running. I've seen the > different options, tcp, http gets, etc, and would like to know anyone > else's experience in what is the best balance over performance and > detecting the lost of service > > Cheers > > Pat > > > [EMAIL PROTECTED] Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=43149&t=43141 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
Re: Content Switching and Keepalives [7:43141]
I see no reason not to use the http keepalive. You can configure it to grab some specific small object if you'd like to limit its impact. We just grab the our homepage and I haven't noticed a performance impact at all. HTH, John >>> "Patrick Donlon" 5/2/02 11:15:19 AM >>> Hi All I have two web servers which are being load balanced behind a CSS, this is working fine. Currently we're using the default ICMP keepalive, this is OK if the failure is at this level but when the web services process is stopped by the DBA the CSS thinks it's up and running. I've seen the different options, tcp, http gets, etc, and would like to know anyone else's experience in what is the best balance over performance and detecting the lost of service Cheers Pat [EMAIL PROTECTED] Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=43147&t=43141 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
