From mboxrd@z Thu Jan 1 00:00:00 1970 From: Michael Tremer <michael.tremer@ipfire.org> To: development@lists.ipfire.org Subject: Re: Test of cleanup branch Date: Wed, 07 Aug 2024 15:50:46 +0100 Message-ID: <F662C347-8A02-4AE0-96E0-595F65AEABE0@ipfire.org> In-Reply-To: <cc70b68c-d3ac-438d-ac1b-03cdefe290cf@ipfire.org> MIME-Version: 1.0 Content-Type: multipart/mixed; boundary="===============3921548830575393026==" List-Id: <development.lists.ipfire.org> --===============3921548830575393026== Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable Hello Adolf, I just pushed a large number of changes and hopefully I have addressed everyt= hing. Please let me know :) -Michael > On 7 Aug 2024, at 13:58, Adolf Belka <adolf.belka(a)ipfire.org> wrote: >=20 > Hi Michael, >=20 > On 07/08/2024 14:19, Michael Tremer wrote: >> Hello, >> Thanks for looking at this again=E2=80=A6 > No problem. This is something that I definitely can help with. >>> On 7 Aug 2024, at 11:56, Adolf Belka <adolf.belka(a)ipfire.org> wrote: >>>=20 >>> Hi Michael, >>>=20 >>> Yes, some of the problems have been fixed. >>>=20 >>> On 06/08/2024 17:52, Michael Tremer wrote: >>>> Hello, >>>> I just pushed a bunch of changes that should hopefully resolve a few of = the problems. >>>> The only one that I can see remaining is that all sorts of form elements= (dropdowns, input boxes, etc.) are now 100% in width. They fill the entire h= orizontal space. This is something I like. However, without the scaffolding a= round being consistent, this creates a lot of problems. Sometimes we have a t= able that is properly sized, sometimes we have I don=E2=80=99t even know what= . I believe cleaning that up will be weeks worth of work because it is so fid= dly. So, maybe I will revert that change and come up with a different solutio= n for OpenVPN. >>> For the DNS Server page you have the protocol entry box which is now the = full width of the page but only has three letter acronyms to be entered. I do= n't have a big problem with that but I suspect that there might be some pushb= ack from some forum members. >> I think I will revert that one change so that we don=E2=80=99t have 1000 n= ew things to deal with... > Sounds a good idea. >>>> Please give the recent changes a test and let me know how it works. >>> I made sure that I did a browser cache clear before reviewing the changes. >> I bumped the version of the CSS sheet so that the browser thinks it is a n= ew file: >> https://git.ipfire.org/?p=3Dipfire-2.x.git;a=3Dcommitdiff;h=3D86ca826ff1= 33eb00b94ae47ac4ab27ac69f037c2 > I ran a CU186 version on my vm and then changed to the latest CU188 version= without doing the browser cache clear and it looks like it did previously wi= th the cache clear so the new css version looks to have worked. >>> The core programs section of the Services page now has colour on it for t= he status of running or stopped. >> Very good. It also has a very different layout. Do you think we need to la= bel the third column in any way to make it clear that it is memory usage? > I am not sure. It seems very obvious to me but not sure about all users. >=20 > Only thing I thought of was to have the first column have the service names= centred. It is similar to the other two columns then and also places the nam= es nearer to the status and memory entries, so is easier to see which one lin= es up with which. >>> The colours are not there for the status of any of the addons. >> That I didn=E2=80=99t touch and clearly break. I did not have any adding i= nstalled on the VM I set up for this. My bad. >>> The Processes and Processes memory graphs that are normally on that page = are now completely missing. >> I removed those. I don=E2=80=99t think that there is any value in them. Th= ey are incomplete, the graphs themselves miss units and legends and labelling= . They either need major work or need to go. Let me know how you feel about i= t. > As you removed them deliberately then that is fine. From my point of view, = I have always wondered exactly how I should interpret what they were showing = to me. I don't think I ever really took much notice of them, so fine that the= y are gone. >>> Screenshot attached. >>>=20 >>> The colour for the status is not showing for the IPSec and OpenVPN client= connections. >> On the index page? > On the index page it is fine. It is on the actual OpenVPN and IPSec pages. = Attached is the OpenVPN screenshot for example. >>> Also if you go to the Firewall Rules creation page the box to select Acce= pt, Drop or Reject has no colour. >> Meh=E2=80=A6 Thank you for finding all these things. I didn=E2=80=99t even= occur to me to look at that. > That is what I am here for. Another colour thing for the top level Firewall= Rules page. Each rule has a line for the actual rule info and a second line = for the remark. If the rule does not have a remark then there is no second li= ne. The alternating grey shading is being done on each line rather than on th= e rule and remark line combined so it ends up looking a bit weird. > Also the colour coding for the drop, reject, accept that used to be at the = start of each rule line is now missing. See attached screenshot. >>> In the Services screenshot you can see that the Intrusion Prevention is s= hown as Stopped but on the IPS page it is showing as running but with no colo= ur. See second screenshot, if you look closely you can see that it shows RUNN= ING for the status. I checked on the command line and Suricata is running so = it is the Services page that has somehow got confused. >> I forgot to cherry-pick the commit that uses the new widget for the IPS. >> I fixed that now and will push shortly. > I will test it when it arrives here. >=20 > I also noticed that for the Samba addon, the top section for the status of = the three daemons does not have any colour coding as it used to have. See att= ached screenshot. >=20 > Regards, > Adolf. >> -Michael >>>=20 >>> Regards, >>>=20 >>> Adolf. >>>> -Michael >>>>> On 3 Aug 2024, at 11:48, Adolf Belka <adolf.belka(a)ipfire.org> wrote: >>>>>=20 >>>>> Hi Michael, >>>>>=20 >>>>> On 02/08/2024 11:12, Michael Tremer wrote: >>>>>> Hello Adolf, >>>>>> Thank you for looking at this in depth :) >>>>>>> On 1 Aug 2024, at 16:01, Adolf Belka <adolf.belka(a)ipfire.org> wrote: >>>>>>>=20 >>>>>>> Hi Michael, >>>>>>>=20 >>>>>>> Additional note written after all of the feedback below the dashed li= ne was written. I did a shutdown of the vm with CU188 to check something in t= he CU186 vm then went back to the CU188 version and things had changed. Most = of the longer boxes are no longer longer and don't overlap other elements. Th= e missing colours are back for IPSec, OpenVPN etc. >>>>>> That is good news! >>>>>=20 >>>>> In some later emails, I indicated that I found that the browser cache w= as the reason that the colours came back. When I cleared my browser cache I w= ent back to no colours and longer overlapping entry boxes. >>>>>>> The graph period selection previously was in a horizontal line under = the graph. Now it was still under the graph but it is now a vertical selectio= n choice which can't have been intended as it takes far too much space. Howev= er the key thing is that rebooting seems to be making things change, which sh= ouldn't be happening. I have started to take screenshots of things to have so= me evidence. >>>>>> I think I have an explanation for this, and no, you are not going craz= y. Since I have changed the CSS, it might be that the web UI is now rendering= some different HTML, but your browser has the old CSS cached and so those th= ings don=E2=80=99t fit together any more. >>>>>> This is a good example: https://git.ipfire.org/?p=3Dipfire-2.x.git;a= =3Dcommitdiff;h=3Dfbfde0088c58c506cab80d23fc240e3cab863302 >>>>>> If you don=E2=80=99t have the changes from the CSS file, then you simp= ly won=E2=80=99t see any colours here. >>>>>=20 >>>>> Yes, I found that the browser cache was affecting things. >>>>> However, when I clear the cache the result I get is the loss of the col= ours in the services table page as an example or the overlapping of entry box= es in the dhcp page as an example. I can get this effect consistently with do= ing a fresh install and a browser cache clear. Have tested this out three tim= es and always the same result. Longer boxes overlapping other elements and lo= ss of status colours. See attached dhcp and services screenshots. >>>>>> Some of the changes I implemented a user feedback - for example that t= he headlines are hard to read on some devices that have not the best font ren= dering. Other things are probably more of my own things. The graphs used to h= ave a grey line around them. And then we had a box with another grey line aro= und them. And then there was the big white box. It kind of reminded me of som= e mirror cabinet. >>>>>> Sometimes we have multiple levels of headlines that all say the same: = Memory information -> Memory Graph -> Memory Usage per Day. >>>>>>> Attached is a screenshot showing the graph period selectors which are= now vertically positioned at the bottom of the graphs. >>>>>> So it is correct that the outer box is gone. The graph only has a head= line when it needs one. >>>>>> The Hour/Day/Week/=E2=80=A6 controls should however remain the same. >>>>> That is definitely different. They are now boxes containing the words a= t the bottom of the graph rather than just the words at the top of the graph = and when first viewed all are grey (graph-example1) then become red when sele= cted (graph-example2). >>>>>>> Also attached is a screenshot of the pakfire page that is still missi= ng the pak_ver numbers. >>>>>> This is probably a bug from my changes in the tooling. >>>>> I have seen the patch. Looks like that will fix it. Will test it to con= firm, either with build and install or using nightly build once done. >>>>>=20 >>>>>>> I don't know what is going on here. >>>>>> Me pushing a lot of changes at a fast pace :) >>>>> :-) >>>>>>> ---------------------------------------------- >>>>>>>=20 >>>>>>> Okay, here is the feedback from reviewing a vm that actually included= the cleanup branch changes. >>>>>>>=20 >>>>>>> The home and ssh items look better with the change. They now use alte= rnating shades of grey to highlight each line. >>>>>> This is now applied automatically to every table. Before, we used to h= ave a lot of Perl magic to make that happen. >>>>> That looks good. >>>>>>> All graphs have new selection buttons for Hour, Day, Weekly etc that = show in red when selected. This looks better. The default is the Day selectio= n but when you start with a graph it does not indicate that it is using the D= ay option, ie all buttons are showing grey. >>>>>> Okay, this would be something that needs fixing then. >>>>>>> Net-Traffic and ExtraHD have the same tables but instead of being cen= tred on the page they are now on the Left Hand Side of the WUI page. >>>>>>>=20 >>>>>>> The Zone Config page looks much busier now. Previously the dropdown b= ox for native or vlan selection was next to the vlan id box and the space was= good enough for that. Now they are placed one above the other, so wider, whi= ch is not needed but they now fill the whole space of the section before the = next nic interface starts. I think the previous version looked clearer. >>>>>>>=20 >>>>>>> The IPS page and the Services page no longer show the red or green ba= ckground for running or stopped. You still see the words but they are now in = white on a light grey background or white on a slightly darker grey. >>>>>>>=20 >>>>>>> On the Services page the first graph which should be titled Processes= is labelled 100%. >>>>>>>=20 >>>>>>> On the System page the graphs have no name at all. On the vm I used t= hey should have been labelled CPU and Load Avg. >>>>>>>=20 >>>>>>> On the Pakfire page the list of available or installed addons do not = show the pak_ver number so that you get alsa-1.2.10- instead of alsa-1.2.10-20 >>>>>>>=20 >>>>>>> The IPSec and OpenVPN pages no longer show the green, red or blue col= ours for the connection status. >>>>>>>=20 >>>>>>> On the Firewall Rules page the table has no colours for Accept, Rejec= t or Drop for any of the rules. The alternate grey shading has got mixed up a= s it is considering that if there is a remark line for a firewall rule that i= s a separate line and the grey shading has to be changed. Definitely not righ= t. Some of the Destinations have not had the right colour code applied for th= e zone colour - left grey. The boxes giving the policy applied for each firew= all section are not colour coded for Allow or Blocked. >>>>>>>=20 >>>>>>> On the actual firewall creation page, there are longer boxes that ove= rlap labels or other elements. There is no colour in the Drop, Reject, Accept= selection box just one shade of grey. >>>>>>>=20 >>>>>>> The vulnerabilities page has the left hand section now in black backg= round with white lettering. Different but probably okay. >>>>>> Okay, this is all very useful. I will install a machine and go through= all of this and see what I can fix. >>>>>>> The following menu items showed longer entry boxes that overlap label= s or other elements >>>>>>> DHCP >>>>>>> DNS Forward >>>>>>> Static Routes >>>>>>> Wake on LAN >>>>>>> Time Server >>>>>>> Log Settings >>>>>>> Proxy Logs >>>>>>>=20 >>>>>>>=20 >>>>>>> The following menu items showed longer entry boxes that did not overl= ap other elements but you end up with a much longer box length than is needed= for a day number for example >>>>>>> Firewall Groups >>>>>>> Firewall Options >>>>>>> Blue Access >>>>>>> Log Summary >>>>>>> Dynamic DNS >>>>>>> Proxy reports >>>>>>> Firewall logs >>>>>>> Firewall graphs - IP, Port & Country >>>>>>> IPS Logs >>>>>>> IP Address blocklist logs >>>>>>> OpenVPN RW logs >>>>>>> URL Filter Logs >>>>>>> System Logs >>>>>>> Captive portal >>>>>>> Connection Scheduler >>>>>>> Assign MAC Address >>>>>>>=20 >>>>>>>=20 >>>>>>> The following menu items showed no change/issues except maybe for the= graph period selection buttons . >>>>>>> Backup >>>>>>> Shutdown >>>>>>> Credits >>>>>>> Mailservice >>>>>>> Memory >>>>>>> Media >>>>>>> Network External, Internal and Other >>>>>>> OpenVPN RW and N2N Statistics >>>>>>> WIO >>>>>>> Hardware graphs >>>>>>> Connections >>>>>>> Mdstat >>>>>>> DNS >>>>>>> Web Proxy >>>>>>> URL Filter >>>>>>> Update Accelerator >>>>>>> Edit Hosts >>>>>>> QOS >>>>>>> IP Address Blocklist >>>>>>> Location Block >>>>>>> IPTables >>>>>>>=20 >>>>>>> I am really sorry for giving false hope that the cleanup branch had g= one really well. >>>>>> Well, it happens :) That is why next is called next. >>>>> Yes but me providing the original first review feedback on a vm system = that did not have the cleanup changes actually in it, really shouldn't have h= appened. >>>>>=20 >>>>> Regards, >>>>> Adolf. >>>>>> -Michael >>>>>>>=20 >>>>>>> Regards, >>>>>>>=20 >>>>>>> Adolf. >>>>>>>=20 >>>>>>> On 01/08/2024 12:50, Adolf Belka wrote: >>>>>>>> Hi Michael, >>>>>>>>=20 >>>>>>>> On 25/07/2024 12:28, Adolf Belka wrote: >>>>>>>>> Hi Michael, >>>>>>>>>=20 >>>>>>>>> On 25/07/2024 10:44, Michael Tremer wrote: >>>>>>>>>> Hello Adolf, >>>>>>>>>>=20 >>>>>>>>>> Thank you for getting back on this so quickly. >>>>>>>>>>=20 >>>>>>>> Maybe I got back too quickly!! >>>>>>>>=20 >>>>>>>> I have just done a build with next for a bugfix patch and installed = it into my vm and the pages look quite different with some things missing, li= ke the green and red colours for if services are running or not and some grap= h titles are incorrect. >>>>>>>>=20 >>>>>>>> I took the iso from the latest directory in the nightly build but it= looks like it was still linked to the previous version when I downloaded it = as the b2 sum is for build 6460dbbf from the day before. >>>>>>>>=20 >>>>>>>> Downloading the iso today from the latest directory gives me the bui= ld ed2c97b7. >>>>>>>>=20 >>>>>>>> I am going to install that version now but I suspect it will show th= e same as I found with my build. >>>>>>>>=20 >>>>>>>> So it looks like my original review was based on the version before = the cleanup branch changes were included. >>>>>>>>=20 >>>>>>>> I will come back with new feedback of what I find from the latest br= anch that I have now downloaded. >>>>>>>>=20 >>>>>>>> Sorry. In future maybe I should wait till the following day, or down= load from the build named directory instead of the directory named latest. >>>>>>>>=20 >>>>>>>> Regards, >>>>>>>>=20 >>>>>>>> Adolf >>>>>>>>=20 >>>>>>>>>> I suppose this also means that the ISO in the next branch boots ju= st fine, too? >>>>>>>>>=20 >>>>>>>>> Yes, that is correct. I used the iso from the nightly next latest d= irectory. No problems with the install at all. >>>>>>>>> This tested out raid and all 4 interfaces. >>>>>>>>>=20 >>>>>>>>> Regards, >>>>>>>>> Adolf. >>>>>>>>>=20 >>>>>>>>>>=20 >>>>>>>>>> Best, >>>>>>>>>> -Michael >>>>>>>>>>=20 >>>>>>>>>>> On 25 Jul 2024, at 09:07, Adolf Belka <adolf.belka(a)ipfire.org> = wrote: >>>>>>>>>>>=20 >>>>>>>>>>> Hi Michael, >>>>>>>>>>>=20 >>>>>>>>>>> I installed CU188 from the nightly onto a vm system and looked th= ough all the WUI pages and everything looked fine. >>>>>>>>>>>=20 >>>>>>>>>>> I then restored a CU187 backup from my testing and then checked t= he OpenVPN RW and N2N. Both worked fine with no problems. Also the logging fo= r that all worked with no problems. >>>>>>>>>>>=20 >>>>>>>>>>> Also checked out the IPS, IP Blocklists, Firewall groups and rule= s, DNS ... >>>>>>>>>>>=20 >>>>>>>>>>> Everything I have looked at shows no impact from the cleanup bran= ch changes. It looks good to me. >>>>>>>>>>>=20 >>>>>>>>>>> Of course good for others to also evaluate. >>>>>>>>>>>=20 >>>>>>>>>>> Regards, >>>>>>>>>>>=20 >>>>>>>>>>> Adolf. >>>>>>>>>>>=20 >>>>>>> <Screenshot_2024-08-01_16-45-00.png><Screenshot_2024-08-01_16-47-11.p= ng> >>>>> <dhcp.png><services.png><graph-example1.png><graph-example2.png> >>> <Services_page_2024-08-07_12-44-08.png><IPS_page_2024-08-07_12-49-33.png> > <OpenVPN_page_2024-08-07_14-39-58.png><Firewall_rules_2024-08-07_14-51-48.p= ng><Samba_page_2024-08-07_14-40-58.png> --===============3921548830575393026==--