|
=== bodie__ is now known as bodie_ |
|
[01:07] <thumper> marcoceppi: *bzzt* wrong |
|
[01:07] <thumper> marcoceppi: juju run runs within a hook context |
|
[01:07] <thumper> marcoceppi: however, it isn't as easy as it should be |
|
[01:08] <thumper> marcoceppi: I managed to execute juju run to get a relation config value using the CLI |
|
[01:09] <thumper> but it requires a bit of prior understanding that isn't entirely obvious |
|
=== zz_CyberJacob is now known as CyberJacob |
|
=== CyberJacob is now known as Guest59898 |
|
=== Guest59898 is now known as zz_Guest59898 |
|
=== mup_ is now known as mup |
|
=== MerryChristmas is now known as benonsotfware |
|
=== benonsotfware is now known as benonsoftware |
|
=== dooferlad_ is now known as dooferlad |
|
[13:48] <pmatulis> with HA, how does the juju master get associated with the mongodb master? is it random (among the initial one + the ones which are about to be created) or is it always the initial single state server? |
|
[13:54] <rick_h_> pmatulis: I'd assume it'd have to be the first ones. Until the others are brought up and sync'd they're not able to be a master. |
|
[13:54] <pmatulis> rick_h_: makes sense |
|
[13:55] <rick_h_> pmatulis: I guess it's not final/official. I think natefinch might know more/better if I'm incorrect. |
|
[13:57] <natefinch> pmatulis, rick_h_: Mongodb chooses a master, and we make that Juju server the master |
|
[13:58] <natefinch> pmatulis, rick_h_: if there's only one state server, obviously it's the master. When you add more, mongodb chooses a new master through its magical and mysterious algorithm |
|
[13:59] <pmatulis> natefinch: ah, the juju master always resides on the same host as the mongodb master? |
|
[13:59] <natefinch> pmatulis: yep, just easier that way |
|
[13:59] <pmatulis> natefinch: perfect |
|
[14:59] <bloodearnest> so, it seems destroy-environment in manual provider doesn't invoke destroy-service on the environments services. Is this expected? |
|
[15:00] <bloodearnest> means I need to remember to manuall destroy-service's first, before tearing down the env |
|
[18:25] <skylerberg> Hi guys. I am starting to write my charm and I see that pretty every charm (that I have looked at) calls execd_preinstall at the start of the install hook. It is not clear to me after looking at the source code for this function what it does. |
|
[18:58] <marcoceppi> skylerberg: this is really for our IS team, they do a special bit that adds an execd directory to call additional files |
|
[18:59] <marcoceppi> it's how they add code for certs and stuff |
|
[19:01] <skylerberg> marcoceppi: Alright. Thanks Marco. |
|
=== zz_Guest59898 is now known as CyberJacob |
|
=== CyberJacob is now known as Guest75471 |
|
=== Guest75471 is now known as CyberJacob |
|
[20:27] <mbruzek> Hello marcoceppi. I am still having trouble getting amulet to deploy with this added binary. I tried deleting the .bzr directory, but then I got juju-deployer errors, and I have tried your commit trick and which didn't work. |
|
[20:28] <mbruzek> marcoceppi: Any other options? |
|
[20:28] <marcoceppi> mbruzek: wait until next week and try again? |
|
|