Given the current discussions about InCommon changing their certs to no longer be functional as client certs, we should look into updating the code to use managed tokens to contact the various services. This may be further complicated by possibly having to use a separate managed token for each experiment. For SAM, that means updating the "samweb_lite.py" code to pass token authentication, reading the token from the appropriate experiments token file; for data_dispatcher and metacat, that means calling the login_token() routine instead of login_x509(), and again getting the token from the appropriate experiment's token file.
For efficiency, we might want a routine that caches the token file text for short periods of time (30 seconds?) so as not to be opening the token file quite so often, but we want to avoid caching it long enough that we would be using an expired one when there is a fresh one on the filesystem.
Given the current discussions about InCommon changing their certs to no longer be functional as client certs, we should look into updating the code to use managed tokens to contact the various services. This may be further complicated by possibly having to use a separate managed token for each experiment. For SAM, that means updating the "samweb_lite.py" code to pass token authentication, reading the token from the appropriate experiments token file; for data_dispatcher and metacat, that means calling the login_token() routine instead of login_x509(), and again getting the token from the appropriate experiment's token file.
For efficiency, we might want a routine that caches the token file text for short periods of time (30 seconds?) so as not to be opening the token file quite so often, but we want to avoid caching it long enough that we would be using an expired one when there is a fresh one on the filesystem.