[01:24:41] --- dev-zero@jabber.org has become available [01:25:05] --- dev-zero@jabber.org has left: offline [04:17:21] --- reuteras has left [04:46:34] --- Simon Wilkinson has left [05:16:15] --- mmeffie has become available [05:25:05] --- SecureEndpoints has left: Replaced by new connection [06:50:51] --- Simon Wilkinson has become available [07:26:30] --- Simon Wilkinson has left [10:22:01] --- SecureEndpoints has become available [11:20:49] --- Russ has become available [12:35:26] --- dev-zero@jabber.org has become available [12:36:20] --- dev-zero@jabber.org has left: offline [13:04:46] --- Simon Wilkinson has become available [13:05:08] "officially OpenAFS isn't supported at Stanford" - really? [13:06:35] --- Simon Wilkinson has left [13:17:17] No. [13:17:35] I followed up. [13:18:18] Mukarram works for a group that threw a completely unfounded hissy fit about how AFS was "insecure" and insisted on removing it from all of their servers while refusing to have a technical conversation about its merits, even after the information security office told them they were wrong, so he's probably getting a different message from his management. [13:20:42] (Mukarram wasn't involved in any of that -- none of it is his fault.) [13:22:40] Also, our AFS documentation is sadly out of date and needs love. [13:38:49] Reeni is also confused about 1.5.x and that it's documented someplace that she needs to incorporate openafs modifications into the kernel (not sure which docs she is looking at) [14:19:49] where is Reeni confused? [14:20:30] has there been a discussion here or in the irc channel that I've missed? [14:21:23] not as of yet [14:21:25] hopefully, soon [14:21:50] I will be on vacation tomorrow to Sunday and tuesday to the following saturday [14:22:01] you have irc and email on your cell phone, right? [14:22:33] yes but vacation == wife does not permit network connectivity [14:22:49] so, wife's in charge :) [14:22:54] I will actually be in the mountains of colorado hiking with my dog [14:23:18] I doubt there will be good connectivity (let alone power) where I'm going [14:23:29] --- Simon Wilkinson has become available [14:23:30] well... have fun, certainly [14:23:41] I plan to [14:23:45] driving or flying? [14:23:49] or train? [14:23:50] flying [14:23:55] then driving [14:24:00] then walking and climbing [14:24:00] the dog is gonna love that :/ [14:24:04] (the flying) [14:24:10] in the cabin [14:24:20] how much extra they making you pay? [14:24:34] --- abo has left [14:24:49] the pet fee is $125 but all the other fees that they are making me pay as a side effect total over $400 [14:24:49] --- abo has become available [14:24:59] RedBear: Are you sure that Reeni isn't just confused about the need to build a kernel module? [14:25:01] side effect of th epet? [14:25:25] and then they reserve the option of refusing to let my pet fly and not issue a refund on the tickets [14:25:36] Simon - no, I'm not... I think the wording of old documentation is confusing her [14:25:46] --- stevenjenkins has left [14:25:56] so, what other related pet fees are they charging? [14:26:40] you must reserve a flight including a pet with a reservation agent. so they charge a per ticket reservation fee (two people each way $25/ticket) [14:27:00] which airline is this? [14:27:31] then they won't let you sit in the standard economy so you must pay the "economy plus" upgrade ($47 each way per ticket) [14:27:44] aroo? [14:27:52] then to check his crate there is a $25 fee [14:27:55] each way of course [14:27:59] of course [14:28:19] I think it might be cheaper to get Leo designated a service animal [14:28:35] then he would fly in the cabin for free and they could not deny him access [14:28:35] --- stevenjenkins has become available [14:28:35] do like Jon and Garfield and dress him up in people clothes... [14:28:48] Odie is a cousin [14:28:56] :) [14:29:47] --- abo has left [14:30:03] --- abo has become available [14:30:50] the only kernel mode change I could imagine that would need to be done for this project would be the implementation of a pioctl if and only if the conclusion is that the query rpc should in fact be implemented through the cache manager. I would argue that this is out of scope for this project but others might disagree with me. [14:31:20] Well, yes, it's intended that it be cheaper for legitimate service animals. But you can't just arbitrarily decide that he's a service animal, and if you can, you shouldn't if it's not actually true, because then you are claiming a special status to which you are not entitled and which you don't really need, which dilutes its value for the people who _do_ need it [14:31:26] GONE [14:32:27] btw, the airline is United. [14:32:31] I think it's the silliness of all the extra fees [14:32:53] back to Reeni - I think we're just discussing building and installing 1.5.x client and server [14:33:27] SecureEndpoints: I would agree with your argument, unless it's easier to implement it through the kernel (I think the query RPC will have to be done with the users tokens) [14:33:31] anyway, I'll let her either find or ask the question [14:33:33] there is no reason she shouldn't be able to build the client and the server including the kernel module [14:33:44] Indeed. I thought she had already built the client. [14:33:45] OK, almost gone. If it's done "through the filesystem", on a path or similar, it should be a pioctl. Eventually. I think it would be fine to declare that not in the scope of the present project. I also think it would not be much more complicated than doing a user-mode client, and possibly less so since you get to take advantage of the CM's infrastructure for managing FS connections. [14:34:08] yeah... I thought you guys had required that as part of the acceptance process [14:34:36] My view is that search isn't done on filesystem paths, initially, but either by fileserver (secureendpoints view) or by volume (my preference) [14:34:57] I also think there's a certain "veil of mystery" surrounding working in kernel code, and more people should be encouraged to look inside and see that it's not really all that different. [14:35:01] --- abo has left [14:35:40] I believe that search scope should be by volume and by cell (its just that we are restricting the model to a single server cell at the moment) [14:35:46] --- abo has become available [14:35:46] Operations on servers and volumes can reasonably be done outside the CM. But then you should probably be extending volint, not fsint, because there is plenty of existing infrastructure for making volint calls from userspace and not much for making fs calls. [14:36:07] jhutz: I would agree about breaking down the veil of mystery. [14:36:13] And now I must go [14:36:18] I have argued that we are building a new service and it should be its own name space [14:36:30] Note that the veil of mystery argument may not apply specifically to Reeni [14:36:38] My worry is that there's only a limited amount of time during SoC, and the less code someone has to grok during that time the better. [14:36:42] You may be correct, in the long term. [14:36:56] SecureEndpoints: I don't care about the name space, but I'd rather not have to use another port. [14:37:33] I think you want another port, long-term. [14:37:42] I'll explain why later, if you don't guess. GONE [14:39:39] *growls at radar map* [16:13:06] --- dev-zero@jabber.org has become available [16:13:12] --- dev-zero@jabber.org has left: offline [16:32:09] --- Simon Wilkinson has left [17:12:07] --- Simon Wilkinson has become available [19:17:17] --- Simon Wilkinson has left [20:42:52] --- mmeffie has left