[08:01] <meena> it looks like OpenBSD and NetBSD have source selection for routes, but FreeBSD doesn't. 
[10:03] <Raghavendra> Hello, I'm trying to use the disk_setup commands in cloud-init but is failing with below errors
[10:03] <Raghavendra> Error: Invalid multi-line string
[10:03] <Raghavendra> On modules/cmx_developer/ldp/main.tf line 113: Quoted strings may not be split over multiple lines. To produce a
[10:03] <Raghavendra> multi-line string, either use the \n escape to represent a newline character or use the "heredoc" multi-line template
[10:03] <Raghavendra> syntax.
[10:03] <Raghavendra> Error: Invalid multi-line string
[10:03] <Raghavendra> On modules/cmx_developer/ldp/main.tf line 114: Quoted strings may not be split over multiple lines. To produce a
[10:03] <Raghavendra> multi-line string, either use the \n escape to represent a newline character or use the "heredoc" multi-line template
[10:03] <Raghavendra> syntax.
[10:03] <Raghavendra> Error: Unterminated template string
[10:03] <Raghavendra> On modules/cmx_developer/ldp/main.tf line 104: No closing marker was found for the string.
[10:06] <Raghavendra> can somebody please check?
[10:07] <Raghavendra> And also we are upgrading Terraform version from 0.11.3 to 1.3.7
[12:28] <aciba[m]> Hello Raghavendra. Could you provide more context, like the user-data you are using? Those files and traces do not belong to cloud-init, so without more info it's hard to know what is going on.
[15:06] <holmanb> meena: good question
[15:06] <holmanb> meena: ip route append is typically for adding multipath routes
[15:06] <holmanb> meena: however that's not what the callsite is supposed to be doing - it's supposed to be adding static routes
[15:07] <holmanb> meena: so I wonder it that is wrong, or not a requirement
[15:11] <holmanb> looks like this commit introduced those calls: https://github.com/canonical/cloud-init/commit/07b17236be5665bb552c7460102bcd07bf8f2be8
[15:11] -ubottu:#cloud-init- Commit 07b1723 in canonical/cloud-init "net: add rfc3442 (classless static routes) to EphemeralDHCP"
[15:16] <holmanb> meena: I need to test this out further, but I strongly suspect that we we can replace the route_append() calls with route_add() and get the same functionality.
[15:17] <holmanb> meena: for example -> `sudo ip -4 route append 2.3.4.5/32  dev enp24s0` yields the same routing table entry as `sudo ip -4 route add 2.3.4.5/32 dev enp24s0` does
[15:56] <minimal> holmanb: the "ip route" manpage doesn't really say much about "append" beyond "With IPv6, further nexthops may be appended to the same route via the ip route append command."
[15:57] <minimal> so its behaviour vs "ip route add" only seems to differ in that scenario
[16:08] <holmanb> right - I'm not sure why it was ever used in the first place since it is barely mentioned in the man page, and ip route add does what we want (and is documented to do so).
[16:09] <holmanb> Given the choice I'd chose the subcommand that is documented to do what we want, rather than the one that is documented to do something else, which also happens to also do the right thing.
[16:10] <minimal> WWCD? (What would ChatGPT Do?) ;-)
[16:11] <holmanb> lol
[16:11] <minimal> as it seems to have become the authority for so many people these days lol
[17:18] <holmanb> looks like a pretty minor change: https://github.com/canonical/cloud-init/pull/2130
[17:18] -ubottu:#cloud-init- Pull 2130 in canonical/cloud-init "net: Use `ip route add` to add static routes" [Open]
[17:30] <meena> holmanb: starting to get the distinct feeling that pull requests are an awful technology for your workflow
[17:31] <holmanb> heh, how do you mean?
[17:35] <meena> see https://jg.gg/2018/09/29/stacked-diffs-versus-pull-requests/
[17:37] <meena> basically: you often have multiple PRs open, each of which is its own thesis. but some of which builds on other PRs. this is really difficult to represent in GitHub
[17:38] <meena> you can't have me independently review and contribute to one of those PRs that build on top of another, because I'm seeing all of it, not just the relevant diffs
[17:47] <holmanb> that's true
[21:31] <blackboxsw> holmanb: I'm good on your dhcp refactor branch. Thanks for the decomposition into separate PRs https://github.com/canonical/cloud-init/pull/2122
[21:31] -ubottu:#cloud-init- Pull 2122 in canonical/cloud-init "[1/2] DHCP: Refactor dhcp client code" [Open]
[22:15] <holmanb> blackboxsw: no problem - thanks for the review!
[22:41] <holmanb> meena: I just merged PR #2122 and rebased #2127 on top of it.
[22:43] <holmanb> that PR has a much cleaner git history now
[22:44] <holmanb> though we still have the PR workflow problem you pointed out, since I'd like to land PR #2130 first
[23:08] <meena> heh
[23:09] <meena> I actually really enjoyed working with BitBucket, when Mercurial was their first Tier product
[23:10] <meena> you could open a PR from a single commit