Register for the iXsystems Community to get an ad-free experience and exclusive discounts in our eBay Store.

SOLVED sub version number really gone, how do you keep the overview with your freenas boxes?

Status
Not open for further replies.

solarisguy

FreeNAS Guru
Joined
Apr 4, 2014
Messages
1,126
Thanks
67
#21
Yes I realise you can downgrade if you have the earlier version. But if I go and install today's, and I want one from 10/12/14, then I should down load the original release and upgrade from that?
And that is not different from realizing that you cannot run 9.2.1.9, and you have to go back to 9.2.1.8 and apply a patch.

Since numbers are easier, I would really prefer to start with 9.3-SU1 and continue with integers. SU followed by a GMT date is good for machines, not humans.
 

9C1 Newbee

FreeNAS Experienced
Joined
Oct 9, 2012
Messages
482
Thanks
47
#22
I'm with depasseg. I have no idea WTF the problem is. I think having the date is a swell idea. Females have trained me to remember and understand the significance of dates. Much easier to relate to than some 9.4.Syphilis.7.1 number. In the end, all we need is a way to tell the updates apart. Although, it could make some great forum topics. "Having iSCSI bugs in gonorrhia.3"
 

jkh

Open Source Developer
Joined
Jul 22, 2013
Messages
1,969
Thanks
1,105
#23
Yes I realise you can downgrade if you have the earlier version. But if I go and install today's, and I want one from 10/12/14, then I should down load the original release and upgrade from that?
This is no different than for any other release. If you want to downgrade all the way to 8.0.4, just grab the relevant installation media and do it. 9.3 does not change the mechanics of downgrading if you are starting fresh. Just what in the heck are you trying to do, anyway? I think you are confusing everyone.
 

rogerh

FreeNAS Guru
Joined
Apr 18, 2014
Messages
1,069
Thanks
118
#24
This is no different than for any other release. If you want to downgrade all the way to 8.0.4, just grab the relevant installation media and do it. 9.3 does not change the mechanics of downgrading if you are starting fresh. Just what in the heck are you trying to do, anyway? I think you are confusing everyone.
Sorry if I am being confusing. I note the new update system raises some new problems which you have mentioned yourself, and I am just putting myself in the position of a newcomer. A couple of days ago the the download link at freenas.org was pointing to the latest stable release.[1] If I had downloaded say 20141214xxxx and had wanted to try (for some reason to do with the latest update) 20141210xxxx I wondered whether I could do a manual 'update' with the 20141210xxxx gui upgrade file, or whether I needed to do a new install with 20141210xxxx, or even a new install with the original 20141209xxxx release and an upgrade. It seems a reasonable question!

[1] Now the download link points to the original release. This may be more and more of a problem as more updates accumulate, unless part of the install routine is to do either successive updates (which would be increasingly tedious as their number increases) or update to the latest SU version. I seriously wonder if you will want to go back to pointing your download link to the most recent update.

Honestly, I am only trying to be helpful and look at the situation from a newcomer POV!
 

jkh

Open Source Developer
Joined
Jul 22, 2013
Messages
1,969
Thanks
1,105
#25
Sorry if I am being confusing. [...]

Honestly, I am only trying to be helpful and look at the situation from a newcomer POV!
But that's precisely what's confusing everyone - you're not! A newcomer will either install the stock 9.3-RELEASE and then, at their option, use the upgrader to catch all the way up, or they will follow the http://download.freenas.org/9.3/latest/ link on the forum or elsewhere. That's it. A newcomer is not going to go out of their way to get any more creative than this (why? What would be the point?). They are also, for the same reason, not going to downgrade to fix a problem - they are going to squawk about the bug and wait for another upgrade. If you find a bug in Windows 7 as a "windows user" do you immediately install Windows Vista? Of course not. This entire premise is flawed.
 

rogerh

FreeNAS Guru
Joined
Apr 18, 2014
Messages
1,069
Thanks
118
#26
OK, I expect you're right. (After all, I'm fairly sure the Windows 8 to 7 downgrading that's going on is sentimental and misconceived.)
 

mjws00

FreeNAS Experienced
Joined
Jul 25, 2014
Messages
798
Thanks
102
#27
Heh. Frankly the 'train' scares the hell out of me, being the paranoid freak robot that I am. I seriously want the masses to ride the train and discover all the pain and suffering long before I integrate anything into production. I'm content with an ISO and a config restore. Inline updating has worked well in my 9.3 testing. But real data gets the jaded sob paranoid prick manual treatment.

This thread pointed me to the wonderful directory with ALL THE DATES!!!! Yee ha. Admittedly I often throw that same string on pretty much anything I am versioning in a flat file system.

The train is love, the train is life. ;)
 
Joined
Mar 25, 2012
Messages
19,154
Thanks
1,849
#28
Honestly, I'm on 9.3 on 2 of my 3 boxes (third isn't migrated because its my main server box and I don't want to break it mid-week). But come Friday or Saturday it's getting upgraded.

Yes, there's been some bugs with 9.3. And they may or may not hit you. I have complete confidence that if I upgrade to 9.3 and something doesn't happen to work for me, I do have the option to go back to 9.2.1.9. I always like to keep my options open. ;)
 

zambanini

FreeNAS Experienced
Joined
Sep 11, 2013
Messages
477
Thanks
32
#29
I started the thread because of the 9.2.1 update nightmare and my update experience with this branch.

august 2013.
Looking forward to replace our redhat storage boxes and the hp iscsi appliances on the long run (to save licence costs on block based replication), we took a look at Truenas.
Truenas sales made clear, that it is not an system like nexenta, even if it is the same hardware, you have to purchase the complete appliance from them.
Ok, sounds fair, since the have to support it. For less important nodes this did not seem to be an option, so use the time with their free product.

So playing around with freenas was great. It looked much more cleaner then napp-it and was easier then vanilla FreeBSD (damn, a cool webinterface which is easy to understand!)
Nexenta was (or still is) in the transition from opensolaris to some BSD distribution.
Open-e decided to release there first ZFS beta (JovianDSS). No more or less HW independent commercial system was around, which will not change completely in the foreseeable following months.

To gain experience, we did a couple of test setups, run our internal tests and decided to deploy a backup node (of a backup system). FreeNAS 9.1 was really charming.
Time went by, everything was ok. So we deployed other freenas nodes for special cases too. I recommended the system to others with no critical usage – everyone was happy.

With 9.2.1 samba had been updated, smb3 was introduced, a few fixes and update releases followed.

Every new version had its own special surprises. AD auth had its issue in one update release, suddenly iscsi stopped, 9.2.1.5 fixed random reboots which did not accure until 9.2.1.3 and so on. It really was a long chain of problems.
Even a update release had been republished.
I do not want to list the issues here, this does not make sense ( and I do not want get through our mantis tickets of that time and the references to the freenas redmine entries: )
So you had to stay on a special release to get sure iscsi, cifs and AD work, the annoying AD Umlaut bug was fixed in the second next update release, but there are bug entries that iscsi has its issues. .-...I guess you get the picture.

The issues with the 9.2.1 branch had it good points too, I took the opportunity to learn more about freeBSD and ZFS.

It seemed to me, ever update release fixed something and something new was broken.
Every „hotfix“ was just untested (in relation to the complete system) and do not be brave to test a new update release until you really have plenty time left to test and go back.
Myself got used to track the redmine entries after every FreeNAS update to get sure, nothing on that non important nodes has an real issue.

Beginning with 9.3 the update train has been introduced. Now it got more complicated to get the oversight and it feels like there is no way back or installing a fresh node with a freenas release and the needed updates. to be sure you are at the release version, you want to be.

it is a little bit like the story tail with mozilla, when the started t tell everybody, that they wanted to stop publishing release numbers.



@jkh, I really do not want to complain.

I really appreciate what ixsystem made out of the original freenas project!
 
Last edited:

jkh

Open Source Developer
Joined
Jul 22, 2013
Messages
1,969
Thanks
1,105
#30
@zambanini - I'm sorry to hear you have had such a rough ride. TBH, none of the issues you raised are a function of release engineering so much as bad QA. Even when we took a really long time between releases and did as much testing as we could, as was the case with 9.2.1.6, things were missed. Scenarios were unanticipated. Some things we just didn't use ourselves and thus didn't realize were broken until someone else ran into them. These are all things that a proper QA department will do for you, but FreeNAS has never had a proper QA department. That is about to change.

TrueNAS, of course, is able to do a lot more QA and QC since it runs on a much more controlled HW environment, release schedules are even longer, and the number of customers (and the amount of things we know about them in advance) are much more tightly controlled. The FreeNAS community is awesome and we wouldn't spend so much time on it if we felt otherwise, but it comes with a lot of challenges and uses some truly and horrendously bad HW (I never thought running a single-drive NAS on a laptop would ever conceivably even be a "thing" until I started hanging out here, for example).

In any case, all of this just begs for a more structured and rigorous QA process that we've never had. We've just hired our first full-time QA resource and are moving forward from there. One of their primary responsibilities will be to create a regression test suite and tell engineering when it's OK to release an update (or major release), this will no longer be engineering's call. It's certainly my hope that this step in the right direction will pay dividends over the coming months and years!
 
Status
Not open for further replies.
Top