Is there a better way to publish to multiple servers
Hello,
We have been using Cascade Server for many years now, and publish to 2 load-balanced servers.
Overnight, we re-publish the entire website, broken into 4 jobs. Our website has 10,000+ pages, and the whole process takes many hours.
We have decided to create a 3rd webserver ( in the cloud ) to improve our high availability.
After adding the third server, third destination and third transport, the publishing jobs are taking about 50% longer. Which I wasn't expecting.
As an example, publishing one particular folder used to take 31 minutes to publish to 2 servers.
After adding the 3rd server, it now takes about 50 minutes.
Based on the log of the job published in Cascade Server, it looks like all the pages are published to one server, then the next, then the next ( see below for a log snippet ). I presume that means each page is rendered 3 times. FTP'ing the files is quite fast, but the rendering takes a long time.
Is there a way to improve this ? Any way to render once and FTP three times ?
Thanks,
Jeff
[Destination: production-server3] langara.bc.ca: continuing-studies/about-us/index.html
[Destination: production-server3] langara.bc.ca: continuing-studies/information-sessions/index.html
[Destination: production-server3] langara.bc.ca: continuing-studies/about-us/learning-units.html
[Destination: production-server3] langara.bc.ca: continuing-studies/about-us/locations.html
[Destination: production-server3] langara.bc.ca: continuing-studies/about-us/getting-here.html
....
[Destination: production-server3] langara.bc.ca: continuing-studies/programs-and-courses/abc-programs.html
[Destination: production-server1] langara.bc.ca: continuing-studies/information-sessions/index.html
[Destination: production-server1] langara.bc.ca: continuing-studies/about-us/index.html
[Destination: production-server1] langara.bc.ca: continuing-studies/about-us/learning-units.html
[Destination: production-server1] langara.bc.ca: continuing-studies/about-us/locations.html
[Destination: production-server1] langara.bc.ca: continuing-studies/about-us/getting-here.html
[Destination: production-server1] langara.bc.ca: continuing-studies/about-us/contact.html
... ...
[Destination: production-server1] langara.bc.ca: continuing-studies/programs-and-courses/index.html
[Destination: production-server2] langara.bc.ca: continuing-studies/about-us/index.html
[Destination: production-server2] langara.bc.ca: continuing-studies/information-sessions/index.html
[Destination: production-server2] langara.bc.ca: continuing-studies/about-us/learning-units.html
[Destination: production-server2] langara.bc.ca: continuing-studies/about-us/locations.html
[Destination: production-server2] langara.bc.ca: continuing-studies/about-us/getting-here.html
[Destination: production-server2] langara.bc.ca: continuing-studies/about-us/contact.html
[Destination: production-server2] langara.bc.ca: continuing-studies/about-us/langara-cs-at-vsb.html
[Destination: production-server2] langara.bc.ca: continuing-studies/events/index.html
Discussions are closed to public comments.
If you need help with Cascade CMS please
start a new discussion.
Keyboard shortcuts
Generic
? | Show this help |
---|---|
ESC | Blurs the current field |
Comment Form
r | Focus the comment reply box |
---|---|
^ + ↩ | Submit the comment |
You can use Command ⌘
instead of Control ^
on Mac
Support Staff 1 Posted by Tim on 24 Mar, 2016 06:07 PM
Hey Jeff,
When publishing very large sites to load balanced web servers, we typically recommend that you do something like the following:
This setup takes the additional publishing to the other nodes out of the equation and will therefore substantially decrease the time it takes for you to update all of your web servers.
Let me know if this helps!
2 Posted by jperreault on 24 Mar, 2016 06:10 PM
Some screenshots of our setup
3 Posted by jperreault on 24 Mar, 2016 06:12 PM
Hi Tim,
Ok, thanks, that does help. With 2 servers it was working fine. With 3 it's becoming a problem. If we ever add more web servers, we're going to have more problems. So, I will look into setting up rsynch. I just wanted to double check that we hadn't mis-configured our transports and destinations.
Thanks,
Jeff
Support Staff 4 Posted by Tim on 24 Mar, 2016 06:22 PM
You bet! Yea, really the only thing you could do to help out at this point (while still publishing from Cascade to all 3 nodes) would be to go through all of your Content Types and just verify that they aren't publishing Outputs to those Destinations that you all no longer use (and if they are, you should disable those particular Outputs). You can also verify that you have Smart Publishing turned on in your Content Preferences as that can save some time as well.
Moving forward, rsync will definitely be the way to go.
Let us know how everything works out!
Thanks
Tim closed this discussion on 24 Mar, 2016 06:22 PM.