Task #8858
Task #8817: Configure sitemaps on the CN
Update CN Apache configs in version control with directives to support sitemaps
0%
Description
Sitemaps are located on disk in ${tomcat_webapps_dir}/${context}/sitemaps as sitemap_index.xml and sitemap%d.xml (for each sub-sitemap).
The rule we've come up with is:
RewriteRule ^/(sitemap.+) /metacat/sitemaps/$1 [R=303]
History
#1 Updated by Bryce Mecum almost 5 years ago
Did a test right now with sandbox and realized this is trickier than I thought. On sandbox, the CN stack (i.e., metacat) is living on a separate VM from the one running Apache. So a hit to https://search-sandbox.test.dataone.org/sitemap_index.xml needs redirect over to cn-sandbox.test.dataone.org/sitemap_index.xml which will result in a broken sitemap. May have to use proxy the request instead of merely rewriting it as we do on the MNs.
#2 Updated by Dave Vieglais almost 5 years ago
If the sitemap entries contain the correct path then ProxyPass and ProxyPassReverse is a simple way to expose the sitemaps.
If the paths need to be adjusted then either correct them on the host or use mod_rewrite for the proxy. Using mod_rewrite will require more resources for the transformation.
An alternative is to rsync them across to search.dataone.org after generation.
Another alternative is to pull them from search.dataone.org and apply a transform on pull.
In all cases, it is necessary to preserve the modified timestamp of the sitemaps so that the correct Last-Modified header is provided with the response. Some harvesters will use the timestamp to determined if further action is required.
Here's a shell script to pull the sitemaps, adjust the URLs and preserve file timestamps: https://gist.github.com/datadavev/8f2ed113bfa16e017a12e0a27f439e5a
For example, run for cn-stage-ucsb-1: https://search-stage.test.dataone.org/sitemaps/sitemap_index.xml
#3 Updated by Bryce Mecum almost 5 years ago
Thanks Dave.
You wrote:
If the sitemap entries contain the correct path then ProxyPass and ProxyPassReverse is a simple way to expose the sitemaps.
This is the case, so I'll go ahead and move forward with a reverse proxy, do some testing, and update back here.
Your script and approach below is slick so thanks for working that up. May come in handy in the future.
#4 Updated by Bryce Mecum almost 5 years ago
Alright, ran a test on STAGE today and this worked nicely.
On MetacatUI host:
ProxyPassMatch "^\/(sitemap.+)" "https://cn-stage.test.dataone.org/$1" ProxyPassReverse "^\/(sitemap.+)" "https://cn-stage.test.dataone.org/$1"
On Tomcat / CN Stack host:
ProxyPassMatch "^\/(sitemap.+)" ajp://localhost:8009/metacat/sitemaps/$1
Note this takes advantage of mod_proxy_ajp and the AJP connector defined in Tomcat's server.xml which I saw already set up and enabled on the CN. It's also my preferred way of running Apache w/ Tomcat these days.
I'm going to coordinate a restart on STAGE with the team sometime soon and discuss the above config and a plan to make the change on search.dataone/cn-ucsb-1.
#5 Updated by Bryce Mecum over 4 years ago
Restart was coordinated last week and things look great on cn-stage.
