Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Testing Sync at Dropbox

dropbox.tech

Post reply on HN
Print view View original

11–20 of 26 posts

Re: Testing Sync at Dropbox

#11

Post by amq »

I'm sure Dropbox is first-class behind the scenes, but the desktop app has become simply awful with time, so that I almost feel sorry for them and embarrassed towards other people when I use Dropbox.

Re: Testing Sync at Dropbox

#12

Post by bluedino »

tasubotadas wrote: post #10
sujayakar wrote: post #2

hey all, author of the previous post ( https://news.ycombinator.com/item?id=22595782 ) here, and I'm happy to answer any questions about the system! I think the author isaac is here too.

Whenever I start Dropbox (boot my laptop) it takes ~30min of maxed out single-core work to "catch-up" since the "last-update". It's just really annoying as it makes my laptop quite loud. Why is that? Google Drive doesn't seem to need to struggle that much.

How much data (total size/count of files) do you have?

Re: Testing Sync at Dropbox

#13

Post by dmd »

tasubotadas wrote: post #10
sujayakar wrote: post #2

hey all, author of the previous post ( https://news.ycombinator.com/item?id=22595782 ) here, and I'm happy to answer any questions about the system! I think the author isaac is here too.

Whenever I start Dropbox (boot my laptop) it takes ~30min of maxed out single-core work to "catch-up" since the "last-update". It's just really annoying as it makes my laptop quite loud. Why is that? Google Drive doesn't seem to need to struggle that much.

So... that’s weird. I have a Macbook Air - not the most powerful machine on the planet - and a nearly full 1 TB dropbox with ~3 million files. Initial resync on boot takes under a minute.

Re: Testing Sync at Dropbox

#14

Post by zmj »

sujayakar wrote: post #2

hey all, author of the previous post ( https://news.ycombinator.com/item?id=22595782 ) here, and I'm happy to answer any questions about the system! I think the author isaac is here too.

Thanks for sharing! I've worked on similar applications - it's always cool to see other successful architectures.

Re: Testing Sync at Dropbox

#16

Post by sujayakar »

tasubotadas wrote: post #10
sujayakar wrote: post #2

hey all, author of the previous post ( https://news.ycombinator.com/item?id=22595782 ) here, and I'm happy to answer any questions about the system! I think the author isaac is here too.

Whenever I start Dropbox (boot my laptop) it takes ~30min of maxed out single-core work to "catch-up" since the "last-update". It's just really annoying as it makes my laptop quite loud. Why is that? Google Drive doesn't seem to need to struggle that much.

hey, can you file a bug from the app? it will collect a report that we can use to diagnose. there's a "Report Bug" option adjacent to "Preferences..."

Re: Testing Sync at Dropbox

#17

Post by sujayakar »

0xy wrote: post #3
sujayakar wrote: post #2

hey all, author of the previous post ( https://news.ycombinator.com/item?id=22595782 ) here, and I'm happy to answer any questions about the system! I think the author isaac is here too.

What are some of the trickiest edge cases that you try to account for? I can only assume that this kind of system is enormously complicated, given the high expectations from users and the complexity of the underlying environments.

here's some of the greatest hits from memory:

1) when we upload a file, we first ask the server which blocks we need to upload. this allows us to deduplicate blocks with other files or previous revisions of files you have. I had written some code like...

  let blocks_needed: HashSet = ...;
  let file_blocks: Vec = ...;
  let to_upload: Vec = file_blocks
    .into_iter()
    .filter(|b| blocks_needed.contains(b))   
    .enumerate();

  for (offset, block_hash) in to_upload {
    ...
  }
the bug here is that we're computing the offset into the file after filtering, so we'd be uploading incorrect contents to the server. we have protections elsewhere that would prevent file corruption, but finding this bug pre-commit with trinity was pretty awesome.

2) nontrivial interactions between different components in the system are hard to test manually and come for free with trinity. trinity will simulate cancelling file transfers, crashing the system, and even dropping the local databases (to simulate disk corruption). these are all hard enough to test in isolation, but knowing how to combine them is almost impossible. so, having "automated test generation" for these cases is really useful.

3) the "theory of trees" seems like it'd be pretty simple, but canopycheck has found some really interesting cases, especially around moves. when we started, we hadn't really thought about how to handle move cycles, concurrent moves that cross ancestor chains, ...

Re: Testing Sync at Dropbox

#18

Post by sujayakar »

OskarS wrote: post #8
sujayakar wrote: post #2

hey all, author of the previous post ( https://news.ycombinator.com/item?id=22595782 ) here, and I'm happy to answer any questions about the system! I think the author isaac is here too.

How smart is the sync engine's "merge" system? Can it cleverly merge files, or if it sees a single file that has diverged, does it just create a conflict copy and let the user sort it out? Like, if both Alice and Bob has changed somefile.txt, but Alice has added a few lines in the beginning, and Bob a few lines at the end (i.e. something that would "git merge" cleanly), will the sync engine merge that into one file?…

yeah, we don't merge file contents. it's really hard to do this well, since you'd need to implement something like operational transforms for every different file type, and the cost of getting wrong is really high, since we'd be corrupting users' files.

I can see us perhaps doing this in the future for more targeted file types where the "rebase" behavior is well-defined, though.

Re: Testing Sync at Dropbox

#19

Post by sujayakar »

saagarjha wrote: post #6

> To cover this layer of our codebase, we also run Trinity in a “native” mode, targeting the platform’s actual filesystem. However, running against the native filesystem incurs a huge performance penalty (roughly 10x), which in turn means Trinity Native can’t test as many different seeds. What platforms do you test in “native” mode? What hardware backs it?

we just use our regular CI infrastructure for running linux, macOS, and windows. we have infrastructure for managing VMs for our different supported platforms, setting up filesystems, and so on.

here's a talk from one of our engineers on our macOS CI infrastructure: https://blog.macstadium.com/blog/virtualizing-mac-infrastruc...

Re: Testing Sync at Dropbox

#20

Post by bmichel »

sujayakar wrote: post #2

hey all, author of the previous post ( https://news.ycombinator.com/item?id=22595782 ) here, and I'm happy to answer any questions about the system! I think the author isaac is here too.

Thanks for sharing your experience on this rewrite. I have a few questions:

1. I'm a bit surprised that you don't persist the scenarios of the tests after a failure, only its seed. Does it mean that when you want to replay it, you have to redo the minimization phase? Or, do you have a way to find a seed for the minimized scenario?

2. Do you have some tests where you generate a set of operations and play them twice: one time with the mocks and one time on the real servers to check that they have the same results?

3. The article says "Note also the importance of the commit hash, as another type of “test input” alongside the seed: if the code changes, the course of execution may change too!". How are you ensuring that a commit really fixes a bug, and not just change the execution path to a happy path where the conditions of the bug are not met? By playing again a lot of tests, or do you write a new unit test that exhibit the bug to ensure the reproductability?

4. Do you think we can say that CanopyCheck is applying randomized testing at the unit tests level and Trinity is applying it at the integration tests level?

Post reply on HN