ironbelly Posted October 11, 2017 Report Share Posted October 11, 2017 Encountered this problem today and it seems completely insane that this is a thing so I'm hoping there is something I am missing or doing wrong. The situation is this: We have multiple servers with our various clients, they are setup as profiles so that when I go into my repo list I see all repos from all servers. A problem with one of those servers cropped up today and the server crashed. The result for me and our entire company is that the entire repo list became unavailable as it was hanging on that one server indefinitely. That one server going down made it impossible to access any repo from the repo list as it would just sit there trying to update forever. Please tell me what I can do to prevent one downed server from making everything inaccessible as it wasn't even clear in the beginning that this was being caused by this one downed server. I had to log in to each and everyone to find which one was offline, reboot it and then check that my repo list came back Link to comment Share on other sites More sharing options...
psantosl Posted October 11, 2017 Report Share Posted October 11, 2017 Hi, You mean listing repos on all GUIs become unavailable? Because obviously clients not using that server can continue working, doing commits... push, pull, I mean, they won't even care if they are not using this crashed server. In case the issue is with the repo list: are you guys often listing repos?? Link to comment Share on other sites More sharing options...
ironbelly Posted October 13, 2017 Author Report Share Posted October 13, 2017 yes this is correct. I was able to continue working on existing repos that had tabs open no problem, but the repo list stayed blank as it got stuck on the downed server when trying to update. In terms of how often we are listing repos, it might be once a week or so, however we onboard new staff regularly so for them it was impossible to get setup fresh on a new project because they couldn't create a workspace as they couldn't access any repos Link to comment Share on other sites More sharing options...
psantosl Posted October 19, 2017 Report Share Posted October 19, 2017 Ok, we'll keep this in mind when reviewing the list repos code. I think it is just looping through them and fails if something goes wrong, silently. Link to comment Share on other sites More sharing options...
SilverKnight Posted March 16, 2018 Report Share Posted March 16, 2018 @ironbelly There is one thing you are possibly missing - not sure if this would be a good work around for you or not (you don't say how many servers you connect to, or how many Plastic SCM profiles you have created) Scenario: I have my own Plastic server at home where I work on my personal projects I also VPN into work where we have a different instance of Plastic - with work projets. I created a "profile" on my work machine specifically to connect to the work Plastic server If I don't start the VPN and just open Plastic for my home stuff, then the repo list fails as you are describing. This is what I've resorted to: I have just installed two instances of plastic scm on my workstation. The normal "install" is used for my home workstation. I then just copied the install folder contents to a new location, then moved the appropriate .config files directly to this new instance. I deleted *all* "profiles" from both instances. The second instance I configured to ONLY connect to my work server, so I can start the VPN before launching this instance. That way, when I open my "home" version, it can see my Plastic server, and doesn't care that it can't get to my work server, because it no longer knows of it's existence (there is no profile configuration for it). *caveats* You now have to update multiple copies of the software (1 for each "server") instead of just running a new install. You also have to ensure you don't overwrite any of the "local"/non-shared configuration files for each instance. You also have to run multiple copies of Plastic GUI when you are working on projects from different servers - so you may need additional RAM to handle the extra instance. Password/Credential changes have to be managed by each of the instances, rather than in the Profile editor - which means you may have to resort to setting creds with the command line configuration tool for each specific instance. If you use an IDE plug-in, this option may not work, because you won't have the correct config fro every workspace) Not an ideal solution, but it does seem to work for me. You can see how to do this from here: 2 clients: (forum post where psnatosl gives the steps) 2 Servers: http://blog.plasticscm.com/2010/12/setting-up-two-plastic-scm.html Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.