Closed cboyden closed 2 years ago
Hi @cboyden! This would be an issue with what data is returned by the Pantheon API. Out of curiosity, do you get truncated output when using Terminus as well?
No, I get the full expected output when I run Terminus. site:list
returns all 1600+ sites. site:info
returns correctly for the sites that don't appear in the select list. And aliases
returns all the sites where I'm on the team.
I just tested this on DDEV with their Pantheon recipe and got the same results. Some sites that I definitely have access to I can't initialize, and others I can. It's the same ones for each platform. If there's not a way to move this issue to Pantheon, feel free to close and I'll file it there.
Hey @cboyden they must be using the same api endpoints that we are 😆
@pirog my guess is that maybe there is some pagination of results going on here or something?
1600+ sites is bonkers, but man, if you have a fleet, you want to be able to Lando it!
@dustinleblanc yeah im not sure either. 270 seems like a rando page size. I wonder how we could even get into a situation to test this?
maybe there is some sort of test org with a shload of sites pantheon could add us to for a bit? maybe we can pass in a pager size so we could handle this with a smaller amount of total sites?
Also might be worth initiating the spirit dance 🔥 🕺 🔥 that summons The @craychee to see if she could point us in the right direction.
@dustinleblanc @pirog When I use --pantheon-site
to initialize a specific site that I know I have access to, Lando must be using the same API endpoint that returns all the sites I theoretically could have access to at once (including all my org's sites). Could it instead use one that returns a single site?
For more background, there's a much smaller number of sites for which I'm on the team. These are the ones I'm most likely to be working on. It's still 200+ sites, though, and I can't imagine ever wanting to press the arrow key 100 times to get to the one that starts with M. So it seems likely that I'd always want to pass in the specific site name, and my colleagues would too. Maybe other users with large numbers of sites would feel similarly.
@cboyden yeah ive definitely heard this specific issue before for people with a lot of sites. they've gotten around it by using the --pantheon-site
flag but im guessing they had sub-270 sites since it worked for them and not for you
Because the pantheon api is not really a public thing i dont really know the answer re: grab a single site. i'm guessing the issue there is we need a site id which we cant get without matching by machine name first.
I think the easiest resolution here is to just get lando to read the whole list of sites since that is the fundamental issue here. Only potential issue there is performance eg how long does it take to grab a list of 1600 sites? what is the best UX around that etc.
@dustinleblanc @pirog you rang?
There have been recent changes on the platform API to the retrieval of sites but the membership API that I know lando hits should not have been impacted. I am happy to dig into this a bit more. Since Pantheon's APIs are not public, I will follow up privately.
@dustinleblanc @pirog @cboyden We've caught a bug on the Pantheon API that perhaps is related. If so, it may have resolved this.
Hey @craychee! Pinged you back via email, thanks again for jumping in!
@craychee @dustinleblanc I am still seeing the same problem when running lando init --recipe pantheon
without the --pantheon-site
parameter: I get a list of 272 sites. (I get 1646 sites from terminus site:list
, which is almost exactly 6 times 272.) I think it might be a different 272 sites, though, because I now see ls
in the list. When I run lando init --recipe pantheon --pantheon-site ls
it works. But when I try a few other sites, I'm getting the error: No such pantheon site!
message.
@dustinleblanc Has there been any movement on the Pantheon API bug? We're stuck on using Lando until we can reliably init all of our Pantheon org sites with it.
@cboyden I haven't touched on this in a while, @pirog do you know where we are with this? I know that we've certainly touched the pantheon.js file in development, but I didn't think we moved the ball on this one at all.
How about this for a workaround: Instead of querying the terminus API for the site, search the user's local cache of sites under ~/.terminus. Perhaps display a notice stating that the user should run terminus site:list
if their site is not found.
Another possible workaround: Get the site ID from the site name using a terminus query terminus site:info sitename --field=id
and then use the returned value to initialize the site.
Also, a manual workaround is to grab a working .lando.yml pantheon recipe file and replace the sitename and site ID with info for the site that's "not found," and then run lando start
. The only problem is that your Lando container doesn't get authenticated to Pantheon. You have to manually authenticate the machine with lando terminus auth:login --machine-token=YOUR_TOKEN
before you can pull the database or files.
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions and please check out this if you are wondering why we auto close issues.
@dustinleblanc this is still a problem.
@cboyden as far as i know this is still an upstream issue with the pantheon api but i know its on their radar so hopefully it can get advanced in the short term
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions and please check out this if you are wondering why we auto close issues.
UNSTALE!
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions and please check out this if you are wondering why we auto close issues.
This is still a problem.
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions and please check out this if you are wondering why we auto close issues.
Unstale please.
On Fri, May 17, 2019 at 7:26 PM stale[bot] notifications@github.com wrote:
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions and please check out this https://github.com/probot/stale#is-closing-stale-issues-really-a-good-idea if you are wondering why we auto close issues.
—
You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/lando/lando/issues/988?email_source=notifications&email_token=AABNJQNMLKBFIWMFCM2Y3PLPV5SNRA5CNFSM4FBNMPB2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODVWF6GQ#issuecomment-493641498, or mute the thread https://github.com/notifications/unsubscribe-auth/AABNJQLIGUJVS7WQOMJCGYTPV5SNRANCNFSM4FBNMPBQ .
-- Brian Wood Application Programmer UC Berkeley - IST Web Platform Services https://web.berkeley.edu
unstale
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions and please check out this if you are wondering why we auto close issues.
unstale
Brian Wood Application Programmer UC Berkeley - IST Web Platform Services https://web.berkeley.edu
On Jul 11, 2019, at 9:23 AM, stale[bot] notifications@github.com wrote:
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions and please check out this https://github.com/probot/stale#is-closing-stale-issues-really-a-good-idea if you are wondering why we auto close issues.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/lando/lando/issues/988?email_source=notifications&email_token=AABNJQLP5RQDZLUWUNYZIR3P65M7RA5CNFSM4FBNMPB2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODZXHQXQ#issuecomment-510556254, or mute the thread https://github.com/notifications/unsubscribe-auth/AABNJQKWOWHLWHD3I3DMTPTP65M7RANCNFSM4FBNMPBQ.
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions and please check out this if you are wondering why we auto close issues.
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions and please check out this if you are wondering why we auto close issues.
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions and please check out this if you are wondering why we auto close issues.
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions and please check out this if you are wondering why we auto close issues.
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions and please check out this if you are wondering why we auto close issues.
unstale
Are there any plans to fix this issue?
DDEV recently fixed their version of this issue: https://github.com/drud/ddev/issues/1232
Alright, i've merged in https://github.com/lando/lando/pull/1997 which bumps the limit here up to 5000
.
Honestly no idea whether that is a reasonable limit or not but guessing its going to be sufficient. This will be in the next release.
Going to close this out for now with the caveat that this change very well may spawn additional issues eg "takes forever to load my site list" so please dont hesistate to open up a new issue if that or something else happens to you!
So to revisit this...5000 sites isn't enough for my org and I need to find another solution. 😞
I'm running into this problem again with the latest version of Lando (3.6.4), but the weird thing is it worked once for a specific site when I initialized inside a script with lando init --pantheon-site "${site}" --recipe pantheon --src pantheon --yes
.
I had to destroy it and re-initialize again for unrelated reasons, and every time afterwards I can't get the site to initialize from the flag, or find it when I walk through lando init
without any predefined options.
Bug Report
Tell us about your setup
v3.0.0-beta.45 on Mac OSX 10.11.6
Tell us about your
.lando.yml
n/a (Bug happens when using
lando init
to create a new .lando.yml)Tell us about the command you were running
Tell us about the error you got
Tell us generally about your bug As a Pantheon org admin, I have access to over 1600 sites. When I run
lando init --recipe pantheon
without specifying a site name, only about 270 of them show up in the select list. When I try to pass in a specific Pantheon sitename, any site that hasn't shown up on the select list doesn't seem to be available. This happens even on sites where I am the user in charge.For the sites that do show up on the select list, the
lando init
command runs fine.