Feedback needed in the sysadm client

[VulcanRidr]

It would be nice to have more positive feedback and messages. Two examples spring to mind. First, I am dual-booting between 12.0 and 10.3. As I was migrating stuff over to /etc, like config files, I went into the boot environment manager and mounted the 10.3 BE. It mounted into /tmp/BE-blah-blah-blah. I cd'ed into the 10.3 /etc, and rooted around for a couple of config files I would need, and verified the firewall settings (I had opened a couple of ports in the firewall config, and I was verifying.

When I went to unmount the BE, the console I had been working in was still in the directory. So when I clicked unmount, a red bar flashed up some text, but it flashed up and was gone too quickly to read the text. The only indication that anything was amiss was that the mountpoint was not cleared. And once I moved out of the directory in the console, the red bar showed up at the top, and the text flashed by too fast to read.

The other place which could use some feedback love is in the Utilities -> Life Preserver. For instance, in the replication tab, I went in and set up replication, but I clicked initialize and start, and other than the button press, got no feedback. Also, in schedules, you can only schedule one snapshot and one scrub per zpool. Say I wanted to schedule, for whatever reason, a scrub every two weeks. The options in sysadm -> LP -> schedules are daily, weekly, and monthly. Given those options, I would have to schedule two monthlies offset by two weeks. But apparently you can't do that. And in the realm of feedback, the add snapshot and add scrub may be grayed out, but with the black and shades of gray of the interface, maybe change the color of the Add buttons to red or something if the option is still available.

One final thing I just noticed was that when you click on the gear to modify scrubs, it resets it back to Daily @22:00, which is a less-than-optimal schedule for scrubs. Shouldn't it at least default to monthly, or default to what the user had previously set?

In general, if a user is doing something right or wrong, I would like to ask that the interface give some feedback on what is going on.

Thanks, --b

[grahamperrin]

https://github.com/trueos/lpreserver/issues/1#issuecomment-250566679 under Life Preserver: replication failed without explanation:

Yea, this particular error looks bad, input/output errors are from the source and will prevent lpreserver from doing anything meaningful.

https://github.com/trueos/lpreserver/issues/1#event-807518808 was closure, I guess that there was no commit (considering the source of the I/O errors) but AFAIR since then with Life Preserver failures for different reasons, I have never observed an alert from SysAdm.

[VulcanRidr]

A couple of more observations, if I might. First, while running KDE, I get two copies of the sysadmin icon in the system tray. I can kill one, but the next time I log in, I have the two.

The other thing that would be nice would be to have the tray icon give more obvious indications of needed activities. Perhaps a priority system, where the icon would have priority levels. For instance a replication failure in the messages might flag the icon yellow (not the pastel yellow that it is currently) whereas a pending upgrade might turn the icon red. Just brainstorming here.

[grahamperrin]

I have been thinking about colours and alerts for a few weeks. Early thoughts:

• no pulse for a momentary normal event, such as a snapshot
• colourless, greyscale pulsing during a scrub
• amber pulsing during replication, because of a bug that is not yet disclosed
• red pulsing after another, more serious bug that is not yet disclosed.

[grahamperrin]

KDE, I get two copies of the sysadmin icon in the system tray. I can kill one, but the next time I log in, I have the two.

That's a separate issue, which was reported or discussed elsewhere. Unfortunately it may be amongst the issues that have disappeared from GitHub :-1:

(I see Bug #15302: sysadmin should present no more than one icon - PCBSD - Bug Tracking System but that's separate from what I recall reading.)

[VulcanRidr]

Another look-and-feel polishing issue. Wen you export public key after generating the certificate, you should have the hostname somewhere in the name.

[RodMyers]

I'd second it. the ability to see what SysAdm is doing is beneficial

[VulcanRidr]

One possible solution would be to have a pop-up when you roll over the icon in the tray, to have it give a run-down of any and all alerts. So the pop-up might have scrub, replication, update ready, etc. but have all listed pertinent ones in that one place.

Another way to improve feedback may be in the area of button click. Instead of clicking on start or initialize in the replication tab, and just having the button turn a darker shade of gray, perhaps have more feedback to tell the user that the command was accepted or rejected. Either a pop-up dialog telling the user, or have the dataset line in question turn green if the command was accepted or red if it was not, with a pop-up telling why.

I'm just spitballing here, coming up with ideas just looking at the screen, and basing it on my experiences.

[RodMyers]

@VulcanRidr , anymore info you would like to pass along?

I was tasked with trying to replicate and/or assign older open tickets

[VulcanRidr]

Here are a few other thoughts on the feedback in the client:

Firewall Manager tab:

• Apparently, not only do you have to add the ports in question not only to the firewall manager in the sysado client, but also to /etc/ipfw.openports
• There should be a way to fill in the used by and description fields in the display. Unless the port is in the Open by Service pull-down, there is no way to fill it in, even updating ipfw.openports.

Service Manager:

• Would be nice if, perhaps a pop-up giving a brief description of the service, either when hovering over the service or highlighting and hitting (or some key combination).

Task Manager

• The only thing here is possibly change "Stop" to "Kill" just to be clear.

Life Preserver: This tab probably needs the most love. Actually, I think zfs replication is the area with which I have had the most problems, so I have probably spent mopre time fighting with this particular interface to get things working.

• Snapshot tab - Name and comment fields should autosize, so that the name and comment are fully visible. There is a lot of empty space to the right. See attached images below.
• Replication tab - The Start and Initialize buttons should be grayed out until a dataset is highlighted.
• Another thought that I had would be to include a way to deploy a different method of doing replication, such as zxfer.

[VulcanRidr]

Another thing that is hit or miss are the logs in the sysadm client. Normally, there are two log entries, except in the rare cases, in which a third entry appears, which is usually Updates Available or Updates Ready.

The other two usually involve snapshots and/or replication. However, they seem to be random and not at all timely. In the past, I have had a problem where a "Replication failed" message from December 16 kept getting dredged up and presented in the sysadm logs through February. I finally managed to clear it by implementing log rotation on lpreserver.log. For instance, in the last 24 hours, I have had the following events:

Jul 19 11:41:59 EDT 2017: Starting replication SCHEDULED task on NCC1764 Jul 19 12:06:44 EDT 2017: Finished replication task on NCC1764 Jul 19 12:32:04 EDT 2017: Starting replication SCHEDULED task on NX74205 Jul 19 13:28:25 EDT 2017: Finished replication task on NX74205 Jul 19 22:00:00 EDT 2017: Creating snapshot on NCC1764 Jul 19 22:00:02 EDT 2017: Starting replication SYNC task on NCC1764 Jul 19 22:00:49 EDT 2017: Finished replication task on NCC1764 Jul 19 23:00:00 EDT 2017: Creating snapshot on NX74205 Jul 19 23:00:01 EDT 2017: Starting replication SYNC task on NX74205 Jul 19 23:00:42 EDT 2017: Finished replication task on NX74205

Now in my sysadm logs, I have the following:

Local system: Updates ready: Restart system to proceed Jul 19 23:00:00 EDT 2017: New snapshot of NX74205 Jul 19 12:32:04 EDT 2017: Starting replication for NX74205

Based on the problems I have had with replication, I would prefer to see replication failed or replication completed. could there be an option for more than two to three entries in the display? Or is it possible to prioritize the messages that get displayed?

[q5sys]

@ZackaryWelch has this been taken care of with all your work on sysadm? Can this be closed or do we need to leave it open?