Repository navigation
Conversation
|
All contributors have signed the CLA ✍️ ✅ |
|
New Issues (6)Checkmarx found the following issues in this Pull Request
Fixed Issues (6)Great job! The following issues were fixed in this Pull Request
Use @Checkmarx to interact with Checkmarx PR Assistant. |
|
I have read the CLA Document and I hereby sign the CLA |
|
This is not addressing performance issues but replacing them with OOME. Might be fine for your use case but with some sites this will OOME in no time. The zap-extensions PR is also introducing issues. |
|
hiya - thanks for taking a look - i tried it on a site with approx 100k pages and didn't experience any OOM - although my dev box has 128GB RAM - but absolutely agree that i don't have a good view on how client machines are configured and what sort of sites are used. i also agree I don't have a good view on scripting and other usage patterns, and hence 'probably not something that can be merged in' - however, maybe it can be useful in identifying some common performance bottlenecks and looking at possible strategies to remove. |
Hi -
Related to #9277 #9278 #9280
Here are a set of changes that addressed, for me, the serious performance issues I found trying to use ZAP to spider a website.
This, and the associated PR that I'll raise in a second in the extensions project - transformed the system from a very slow thing with an unresponsive GUI to something that was very fast, and GUI was responsive throughout.
I appreciate that this PR is probably not something that can be merged in. However, I'm sorry that I don't have the time to split it, and shepherd it through a review process. However, it is offered as-is if it is useful.
The main changes are:
ParosTable* -> these classes were very poorly performing as they were used by many threads at one time, but were synchronized at the class level, meaning very heavy contention. Additionally, each write was followed by a read from an hsql database - which was quite slow. Certain of them have been replaced by an InMemory* version - these appear to have the same behaviour as the existing classes, but have not been verified extensively. It is possible that in some uses cases they are no longer suitable.
Some methods in ParosTable* which appeared to be unused were removed. This was to understand which methods needed to be implemented in the InMemory* versions - these deletions had no effect in my use of the system. it is possible they are used from scripts?