| 1 |
Am Tue, 21 Mar 2017 23:14:16 +0200 |
| 2 |
schrieb Alan McKinnon <alan.mckinnon@×××××.com>: |
| 3 |
|
| 4 |
> On 21/03/2017 22:25, Kai Krakow wrote: |
| 5 |
> [...] |
| 6 |
> >> Good idea. For posterity so's I don;t forget: |
| 7 |
> >> |
| 8 |
> >> - gentoo home page |
| 9 |
> >> - one or more readthedocs sites (especially docs.ansible.com) |
| 10 |
> >> - 1 or more jira pages (work stuff) |
| 11 |
> >> - 1 or more confluence pages (work stuff) |
| 12 |
> >> - a local gitlab site |
| 13 |
> >> - 1 or more python html apps for, errr, usenets stuffs |
| 14 |
> >> - 1 or more python apps running in flask (work stuff) |
| 15 |
> > I'm using Chrome with often more than 50 tabs open - without |
| 16 |
> > performance problems. But I had to switch Chrome to simple HTTP |
| 17 |
> > cache (chrome://flags/#enable-simple-cache-backend) to get that. |
| 18 |
> > Some tabs take more than 1 GB of memory after a while. But only |
| 19 |
> > those tab become sluggish, all others are unaffected. And I do not |
| 20 |
> > block any scripts (just the usual performance killers like ads and |
| 21 |
> > tracking scripts, done with Ghostery). I cannot believe that |
| 22 |
> > Firefox is so much worse at this? |
| 23 |
> |
| 24 |
> Oh, I can believe it all right. Firefox is a bloated monster. The |
| 25 |
> devs keep refactoring it to get rid of bloat, but it's a lot like |
| 26 |
> emptying a lake with a teacup. |
| 27 |
> |
| 28 |
> same with bind, open|libreoffice, bash, gcc ... |
| 29 |
> |
| 30 |
> > If Chrome isn't a option for you, maybe try Vivaldi (it's based on |
| 31 |
> > Chromium and said to be very fast, tho for me it starts much slower |
| 32 |
> > than Chrome). |
| 33 |
> |
| 34 |
> Using Chrome isn't really a problem. I can use anything I feel like |
| 35 |
> (no overlord here!. More correctly: the overlords that do exist are |
| 36 |
> ineffectual when confronted with a determine sysadmin) |
| 37 |
> |
| 38 |
> The problem is that this 52 year old is tired of learning knew |
| 39 |
> workflow and retraining muscle memory for reasons of just because. |
| 40 |
> He's happy to learn python (useful), learn ansible (very useful) but |
| 41 |
> not chrome as the everyday browser (meh) |
| 42 |
|
| 43 |
Well, either arrange with the one or the other options. There's not |
| 44 |
many more options... ;-) |
| 45 |
|
| 46 |
Working in a environment where I'm a system deployer, level-1 and |
| 47 |
level-2 supporter, network and sys admin, all in one person, I know how |
| 48 |
difficult it is for a human to learn new stuff. |
| 49 |
|
| 50 |
So one part of deployment is to keep things simple: Don't bloat the |
| 51 |
browsers, as one example. Chrome can be very minimalist. Actually, |
| 52 |
there not much visual difference between browsers today. I always unify |
| 53 |
that by eliminating the visual differences: Remove the search bar from |
| 54 |
firefox, rearrange some buttons, and it looks like almost any browser. |
| 55 |
Unify the function differences by making the search function into the |
| 56 |
URL bar (maybe with some addon). Well, but that's too late now. ;-) |
| 57 |
|
| 58 |
So the other option is get rid of the stalls. First, try with an empty |
| 59 |
profile and see if the problems persists, improves or gets worse. |
| 60 |
Depending on that we could try finding the correct direction for next |
| 61 |
steps. |
| 62 |
|
| 63 |
Usually, it's either blocking on IO (an issue to be fixed in system |
| 64 |
configuration), or it's blocking on CPU (an issue with plugins/addons |
| 65 |
usually as the Firefox core should be designed to work mostly |
| 66 |
non-blocking these days). |
| 67 |
|
| 68 |
The following tools may become your friends: top and iotop. |
| 69 |
|
| 70 |
-- |
| 71 |
Regards, |
| 72 |
Kai |
| 73 |
|
| 74 |
Replies to list-only preferred. |
Archives