Big Spikes in egress

Or load tests in production :wink:

IT’S BACK!!! lol

Interesting, I also see spike, but small (west coast)

Looks like a big customer moving somewhere else. Getting a lot of trash and storjstats.info also reporting a decrease in used space.

I’m not really seeing any more trash than normal at the moment… but well see.

1 Like

They may also just use egress. You can easily to check that - hover mouse above graph and you will see is it egress or repair. If repair - it perhaps moving to geofenced nodes, if egress - it is a regular usage.
The deletions happen every day, I did not see any spikes in that, except when EUN-1 and US2 were decommissioned: Announcement: Storj to shut down europe-north-1 and us2

Getting hit with an egress spike here today as well.

Thank you for clarifying.

I also support reasonable general skepticism but not questioning the integrity of individuals who have proven themselves reliable.

Enjoy your egress spikes!

These are not peaks, they are the return of normal egress. I hope the winter is ending soon and spring is coming into our nodes. :money_mouth_face:

2 Likes

Sadly my busiest node is still going through withheld so I won’t get to keep most of what it earns at the moment.

1 Like

Yeah its back. Super great! :money_mouth_face: :yum:

My egress is all from cache… almost nothing of it is from actually storage reads…
but i do have a lot of cache…

maybe we are streaming something popular

my reads from storage is like 5% and 95% from RAM / Cache (meaning repeat business)

4 Likes

Do you mean L2/L3-Cache? I have lots of RAM and the new Intel Core i7-13700K running. And it has lots of L2/L3-Cache.

my server got 224GB RAM so ZFS ARC gets 112GB, and zfs is pretty smart at keeping track of what is most efficient to keep in the ARC…

so yeah cache, but storage cache, not CPU cache

Ooooh, nice specs!
Wish I could play with a toy like that :wink:

What exactly is ZFS ARC? Am running all the nodes on EXT4.

How many nodes do you have running on this 224 GB Machine?

It has 54 nodes, two 24 Bay disk shelves and 12 bays in the server itself.

my node has only 2,5GB of RAM, and i observed constant upload over minutes, on level i didn’t saw earlier, but no constant HDD reading, means there has to be some parts read once and keept in RAM i guess too? and send constantly, i have a simple win 10 and NTFS HDDs. it can also be synthetic help, due to locally low egress last weeks, mayby they want help SNOs finnaly, ¯\(ツ)/¯.

1 Like

yeah i duno whats going on…

it could be that Storj wants to actually give SNOs a bigger payout, while incentivizing better egress internet speeds for SNOs.

does seem a bit weird.

:money_mouth_face: :yum: