Globus

h1. FAQ

h2. I'm unable to link my Globus account to my NCCAT/MEMC account from the CUIMC Campus

The typical advice we give to Columbia affiliates if they encounter an error with Globus account linking is to first switch to a different network (e.g., a WiFi hotspot).  This is because our Globus endpoint domain (_globus3.5540bd.75bc.data.globus.org_) apparently cannot be used on certain CUIMC networks.  Specifically, our Globus domain cannot be resolved (i.e., its domain name cannot be mapped to an IP address).  The _domain globus3.5540bd.75bc.data.globus.org_ is managed by Globus (not us) and in turn, uses Amazon as a domain name provider.  Ultimately this means that there is a problem with CUIMC IT's domain name infrastructure where Amazon domains cannot be reached; from the outside it's unclear if this an actual issue or just a matter of policy related to the protection of patient data / PHI.

If you keep encountering this issue, our advice would be to contact CUIMC IT and/or Globus support.

h2. How do I unlink my MEMC account and link my NCCAT account (and vice versa)?

If you currently have a MEMC account linked to your Globus account, you may not be able to access your NCCAT directory (or vice versa). If this is the case, you should use the following instructions to unlink your MEMC account and link to your NCCAT account:

1. Click on _Manage Identities_ on the _Settings_ page in the Globus web UI.

2. Then on the next page click on the trash can icon in the same row as _<username>@globus3.5540bd.75bc.data.globus.org_ and then click on _Unlink Identity_.

3. Next time you are prompted to log in with your credentials, use your NCCAT account and its associated password.

h2. My Globus transfer times out every 5-10 minutes and is proceeding very slowly

As an initial troubleshooting step, we suggest that you run your transfer over a VPN (such as ProtonVPN or Mullvad).  If this does not resolve the issue, please contact the MEMC or NCCAT office and we can assist.  A longer explanation of the most common cause for this issue can be found below.

If your Globus endpoint is on a university/research institute campus, the particular issue that you are encountering is most likely caused by a quirk of how internet routing works.  Our network engineer has configured our gateway router so that outgoing traffic is sent over "NYSERNet":https://nysernet.org/home (our regional research and education network).  Our network also sends a message to other networks  (such as your institution's network) telling them to send incoming traffic over NYSERNet as well.  However, sometimes this message (a route advertisement) does not reach a network, and incoming traffic is routed differently.

This asymmetry in routing (outbound traffic over NYSERNet, inbound traffic from another network) causes the connection to become periodically interrupted, and every time this happens your Globus client needs to create an entirely new connection, which slows the transfer down considerably.

Our network engineer's workaround for this is to manually implement an exception to our normal routing logic.  For this to work, we require a fixed IP address or range for the destination that traffic should go to.