Closed melrom closed 10 years ago
Shantenu is handling with Yaakoub and XSEDE admins.
Service is up. Users need to be informed.
Melissa -- can this be closed? What's the status of redis.tacc ... Is it production ready?
Closing.
Here is our status: our users can use gw68 permanently. Depending if we resolve TACC firewall issues, we have a redis server running on repex1. We also have redis01.tacc... however Ashley has said she needs this one for BigJob performance testing, therefore it's unstable (i.e. if she slams it with O(100000) jobs, it might crash) and she also periodically has to clear it. Wouldn't recommend that to users.
We should follow up with the suggestion by Dave Wright's group for hosting one over on their side of the Atlantic.
I'd be happy to create a redis on repex1 for Ashley's experiments, so that we can use redis01 for user backup...
A.
Currently, we need to inform users of the need to use their own redis server to run BigJob. We also need to come to a concensus on the easiest way for users to start a redis server to do the tutorials.
Andre Merzky proposes a solution in an email entitled "REDIS redux"
To quote his email:
I wrote a small server which does exactly that: it creates redis instances on demand and returns its port and password. That instance is then killed after a certain time (by default 24 hours, but can be extended at will).
It should be quite easy to include that in the tutorials. For example:
echo "REDIS CREATE" | netcat -q 2 localhost 2000
201 creating redis instance '299ea641-94e0-4a2c-902b-f7000d8fd2b9': redis://localhost:10000/
gives a unique redis instance -- and it gets cleaned up after the tutorial. The instances are independent from each other, and as secure as any other redis installation (AFAICS).