Jump to content

wnicholls

Members
  • Content count

    30
  • Joined

  • Last visited

Community Reputation

0 Neutral

About wnicholls

  • Rank
    Advanced Member

Profile Information

  • Gender
    Not Telling

Recent Profile Visitors

1,732 profile views
  1. Upgrading to 6.0

    I upgraded a server from 5.4 to 6.0 just fine without any issues. I'm usually pretty gung-ho about Plastic upgrades, (relying on having a good backup system in case something _does_ go wrong) but for me every time is faultless. I can't really tell you what 6.0 with SQL is like, because I did switch to Jet pretty soon after the upgrade. This is an aside, but as far as faultless goes, I can't say the same for the migration from SQL Server to Jet though. The main issue here is that it has to be done per server and all repositories have to be done at once. Codice's own recommendations probably are along the lines of "set up a second server, and migrate all the repos over, then when you're happy, turn the first server off" and although I wouldn't go that far, I would definitely recommend (a) Get all users to check in their work (b) back up (c) backup again! (d) get the users to back up their workspaces; and (e) test very carefully after the migration. Reverting back to SQL I *think* is just a case of changing the db.conf file back. (I'm sure I did that when, back in 5.x days, I tried to replace MS Sql with MySQL. That *was* a complete disaster for me, but for posterity/observers, I'll just say Plastic apparently works just fine with MySQL for anyone else). I had one repository where a changeset seemed to have vanished after the conversion to Jet, leaving a workspace in very strange state. I don't know the reason for it, everything else is ok, so doubt that will be a very common case, but there isn't that much experience of migrations to Jet out there yet. I recovered by taking a copy of the workspace, deleting and recreating it and updated to get the latest version the _server_ thought there was, copying the original workspace files over the top, and checking in again. You can imagine how scary that would have been if the I didn't have all the changes somewhere.
  2. Please bring back the workspace tabs

    Tabs are a well-understood concept - but NESTED tabs? Having two different widgets for two different purposes helps separate them. Everyone's use case is going to be different but personally I am comfortable with the 6.0 drop down. On the other hand, I've found the way global views - workspaces, repositories - mix in with the things like 'items in workspace' , to be really stupid: This clutters up the workspace view for starters, and with multiple workspaces they get lost. The obvious way is for this to be some kind of "Global views" set of tabs. Not sure if that will be significantly more usable - don't just take my opinion. In some respects they are really singleton views - why have more than one open? - and should behave like Preferences does - I mean here not so much the popup window, but that clicking on "Workspaces" in the sidebar would bring up THE workspaces view. Of course, other people might like to have multiple views with different filters set. Not that you can really do that successfully now.
  3. Suddenly all the changeset is missing ?

    Case in point - I just opened Plastic GUI this morning, on the workspace I had open yesterday, but the filter settings have reset to today's date and it now only shows the one change set in the last month.
  4. Suddenly all the changeset is missing ?

    A "select top 100 " type of filter would be a much more useful default. I'd probably never change it from that. I operate a small custom software development company. Customers can go for months to years between changes being made to their software, so this is a very common scenario for me: (1) open up a repository that hasn't been touched in several months (2) go to see the history of the project - more often than not, the changeset view is empty at this point. I then have to make a wild guess as to what date to put in to the filter to get to see something. Even if the repository is YEARS old, the default view of the change set should show something. Without experimenting and thinking through, I'm not sure what the consequences would be, but certainly when you create a workspace for the first time, you can set the filter to whatever you like, and actually if a literal date is remembered now, then as time goes on with an active project, the list of change sets is going to get longer and longer yet the developers are usually only going to be interested in the top few.
  5. Thanks - I thought I had put a comment on that thread but can't see it now (perhaps I got as far as preview). At time time it seemed the symptoms were different enough to deserve a separate thread - the first post on that thread is nothing like my issue, Dinesh's is exactly. Note I don't seem to get anythng logged in webui.log.conf though, unlike rodger.
  6. Executive summary: WebUI install on Windows Server 2016. Browser gets login window, however no user can log in - browser shows "Login failed.The user name or password provided is incorrect." - webui.log.txt shows " ERROR Authentication: Unable to validate user 'USER': User / password has not been configured. Please run the configuration wizard to set them up." Much longer information: Windows Server 2016 standard. Server's real name is Playpen, but host name for PlasticSCM purposes is "plastic.csl" (can't be problem, but so you understand any logs / configs posted here). When I started attempting to configuring Web UI I was on plastic v 5.something and first halfway-successful attempt where I got these symptoms was with version 6.0.15.1158. But upgrading made no difference. Plastic Admin Tool shows "Version 6.0.16.1365 " and Authentication mode "Plastic SCM own user and password database" Licence expires 27/10/2017 (many months away). Database backend - Jet./ Listening on port 8087 (with SSL on 8088 although this isn't relevant). Server Configuration wizard shows all the same stuff. Plastic "Users and Groups management" tool shows two users, "walter" and "matt" Web UI frrom PlasticSCM-6.0.16.1365-web-frontend.zip was installed , into root for Site http://plastic.csl/ it runs in its own Application pool with the identity set to "LocalSystem" NO matter which user attempts to log in through browser, get the same "Login failed" message web.config contains - I've confirmed this is the active one ( by making a visible change that did take effect) <appSettings> <add key="Language" value="en"/> <add key="SEIDWorkingMode" value="UPWorkingMode"/> <add key="Server" value="localhost:8087"/> <add key="IdleProcessUser" value="walter"/> <add key="IdleProcessPassword" value="--deleted here but was the right one--"/> <add key="LogRequestTime" value="false"/> </appSettings> ( Out of desperation at one point I trying changing "localhost:8087" to "plastic.csl:8087" which is how all the developer machines are configured, but as expected, no improvement ) webui.log.conf contains: <logger name="Performance"> <level value="DEBUG" /> ... Global = DEBUG, Authentication=DEBUG and <root> level = DEBUG webui.log.txt shows: 2017-07-13 16:48:48,437 INFO Global: Application started 2017-07-13 16:50:37,645 ERROR Authentication: Unable to validate user 'matt': User / password has not been configured. Please run the configuration wizard to set them up. That's all it shows. No Debug information of any kind. No matter what I fiddle with, no change in symptoms, and no further debug info in this log, either. plastic.server.log shows this . @17:03:29 is the server starting up ... 2017-07-13 17:03:29,412 (null) NT AUTHORITY\SYSTEM at (null) INFO Channel - Using certificate file C:\Program Files\PlasticSCM5\server\ssl-certificate.pfx 2017-07-13 17:03:29,663 (null) NT AUTHORITY\SYSTEM at (null) INFO Channel - Start listening on protocol ssl:// port 8088 2017-07-13 17:03:29,667 (null) NT AUTHORITY\SYSTEM at (null) WARN Tube - No license for Tube, won't initialize. @17:04 I visit http://plastic.csl and are eventually presented with login page. This log entry probably has nothing to do with that! 2017-07-13 17:04:31,129 NT AUTHORITY\SYSTEM at INFO Transaction - Transaction timeout -> 120000ms @17:05 I attempt to login as "walter". Nothing at all is recorded in the server log. (webui.log.txt shows yet another Unable to validate user 'walter': User / password has not been configured. " Server log did have activity like this previously which is the Windows GUI running on a remote machine .. 2017-07-13 16:49:26,035 NT AUTHORITY\SYSTEM at INFO ChannelCall - protocol:plasticproto recb: 2|rect: 0|sentb: 88|sendt: 0|queuedt: 0|prt: 0|th: 14|dest: 0|mt: 0|sert: 0|zip: 0|cpu: 0| 192.168.68.44|user::walter|GetLiceseData 2017-07-13 16:49:54,265 NT AUTHORITY\SYSTEM at INFO Security - AuthPerCall running. bDontCheckAuth = [False] 2017-07-13 16:49:54,275 00000000-0000-0000-0000-000000000000 NT AUTHORITY\SYSTEM at PLAYPEN INFO Operations - Retrieve repository list 2017-07-13 16:49:54,428 NT AUTHORITY\SYSTEM at INFO ChannelCall - protocol:plasticproto recb: 2|rect: 0|sentb: 1128|sendt: 0|queuedt: 16|prt: 250|th: 14|dest: 0|mt: 0|sert: 0|zip: 0|cpu: 93| 192.168.68.44|user::walter|GetRepositoryList So - I'm right out of ideas. Does anyone else have any ? - Walter
  7. I think I can see why "plasticscm-latest" worked for jonask and "common" did not. Neither work for me today because the keys had expired. Here's the an excerpt from running apt-key list on my machine (*buntu 17.04) pub rsa2048 2015-01-12 [SC] [expired: 2017-03-22] uid [ expired] plasticscm-common OBS Project <plasticscm-common@obs2.codicefactory.com> pub rsa2048 2015-02-13 [SC] [expired: 2017-04-23] uid [ expired] plasticscm-latest OBS Project <plasticscm-latest@obs2.codicefactory.com> Top one is a red herring. Other two dates are the two plasticscm keys. Note latest expiring later. Fix with: wget https://www.plasticscm.com/plasticrepo/plasticscm-common/Ubuntu_14.04/Release.key -O - | sudo apt-key add - wget https://www.plasticscm.com/plasticrepo/plasticscm-common/Ubuntu_14.04/Release.key -O - | sudo apt-key add - These update keys won't expire until June 2019.
  8. Long delay on this thread, but since I hit this today (for the second time in two days). It would be useful for the pending changes delete to ALSO have the option - by default - of not removing files from disk. It strikes me as pending changes is more likely the place you would spot this problem - eg a build output directory or cache or something acciddently added at some time, then you find the problem when files in it start changing... Probably should log this in uservoice : I've put it there
  9. Great great news. Meanwhile the instructions at https://www.plasticscm.com/plastic-for-linux/ubuntu-16.html allowed me to get gtkplastic working on Ubuntu 16.10, with a few minor issues. Of course ubuntu is now 17.04, I'm sure you'll test with that.
  10. Z_DATA_ERROR when updating workspace after migration

    For a follow-up to my end. The problem was clearly related to conversion from SQL Server to SQL CE last year -for whatever reason, the repository didn't convert properly. This was not noticed immediately because everyone's working copies were still ok and so were new commits. However any access to the corrupted change sets (such as from a history or diff, or attempting to create a new workspace) was finding the errors. Solved for me by taking the original SQL Server backup (before migration) then replicating the changesets since. I suggest anyone else with Z_DATA_ERROR should log a ticket with support since it isn't clear there is only one cause and the solutions will depend on your situation. THanks for your help Carlos,
  11. Z_DATA_ERROR when updating workspace after migration

    Sorry to reopen an old thread but guess what - happened to me. Brand-new Windows 10 machine. Brand-new Plastic SCM client install - version . Authenticate to plastic server ok. Creating very first workspace (by joining existing project). First Update and... 8 files are listed with "Z_DATA_ERROR" out of perhaps 40 that should be there (plus subdirectories). So what I tried was copying the 8 files from a good version of the source, then in Pending changes, they were listed as "Added and Pirvate items - 8 of 8 items selected". But I can't add from there - get error "Can't add an entry with the same name. Duplicated child [filename]. Parent []" So the data in the repository is presumably damaged. But how do I fix it? Well... I tried this and a few weird things happened. Delete all the files from the workspace again. Now under 'pending changes' they come up as 8 of 8 Deleted items. Check in the delete. Now copy the good files back in. They come up as "added and private" Attempt to check in. Same problem "Can't add an entry with the same name..." But now, when I attempt to update the workspace, only two files are coming up with Z_DATA_ERROR. The other files seem to be around. Worse the changeset that should have deleted them never appears to have made the database. So presumably the bad ones are still in there. I deleted the workspace and recreated - back to the bad 8 files. So indeed, my attempts to fix don't work. I tried another workspace- similar era of repository creation. Same problems. But yet another repository is fine. This may explain some problems I've had with annotate and diff operations failing, which I was suspicious had to do with file type or file size+file type but now probably have to do with bad repository data. Now I have good copies of all the files that are damaged, so I if I could find a way to force them into the repository, but ... how? Thanks Walter
  12. Thanks Carlos. I got around it for now. Sounds like best we just wait as patiently as we can because when you do fix it, it will be properly fixed. P.S. I am in no position to criticise timing - in something completely unrelated to this, I've just finally attended to a problem that took me 30 seconds to fix, but the customer has been waiting for three days... also note how long it took me to response on this thread!
  13. Should have mentioned that I am now explaining to a client why their project is going to be late, because I am unable to create a clean build machine. Well, I can create a build machine just fine, but I can't do a simple pull to get the source code onto it!
  14. I don't wish to sound ungrateful, but 3 months have gone by, From another thread (from manu): "it's in our roadmap to support the new apt-get version ... As you said, installing Plastic SCM using the binaries over an existent installation should work just fine." What about installing the binaries over a non-existent installation? ie a new install. I don't wish to sound ungrateful, but linux support in PlasticSCM had historically been very good, but now as far as I can tell, it is impossible to install Plastic SCM on any flavour of Ubuntu from 16.04 onwards, using any of the instructions that Codice have supplied, or for that matter anywhere else on the internet that I can find. As a paying customer - and aren't all your customers paying - we do deserve better than this. In fact, decent linux support is one of the things that led me to choose Plastic over other (the other being a really good GUI)> It doesn't strike me that getting SHA512 keys on the repository would be all that much work, assuming that was the only steps required - but the Mono versions in question are available in the distro's standard software stack. - Walter
  15. Still haven't had time to investigate much but "An unexpected error has occurred" even when running sudo. At this error it prints "Marshaling changed signal" to stdout but no log written that I can find. in the scripts directory is a gtkplastic bash script, these are non-functional as provided, but if I do my best by chmod+x and moving around, I can't make them do any better. I don't have time to hunt around (inadequate) instructions and debug your code etc etc ..... so would muchly appreciate you signing the repository with a SHA512 key .. better still issue a proper 16.04 LTS package. Thanks in advance.
×