Existence of switch disturbs network: cable is fine straight to laptop but not through a switch

Solution 1:

As you've already noticed, chaining Ethernet switches is not a good idea. There may be limitations in the cabling but even that should be corrected some day.

Generally, Ethernet switches should be arranged in a tree (or multi-star if you like): a central switch and all other switches connected to that central switch. This topology has two main advantages: if any switch but the central one dies, just a small part of your network is down (same goes for the cabling itself), and each switch gets the full bandwidth of its uplink port (rather than sharing that across a chain).

Connecting two switches to each other only works if at least one of them supports Auto MDI-X, otherwise a crossover cable is required. Auto MDI-X is pretty much standard today but not ubiquitous. It seems to be supported by the TL-SF1005D, however[*1]. On the HP 2530-24G it's active by default but can be deactivated. Check whether the port LEDs light up when plugging the cable. If they don't light up a crossover is likely required.

Another possibility for connectivity loss is a network loop. I'm not sure that can be ruled out from your description, but if more than the connection at hand is impacted it's almost certainly a loop - do you see the traffic LED go solid?

The 2530 supports both spanning tree protocol RSTP/MSTP and loop detection - both are off by default. RSTP/MSTP logically blocks a detected loop, you won't see the port lights go off! Loop detection deactivates the port altogether for a configured time period and the port lights will go dark. Check the 2530's config to make sure (show running-config). Also check the 2530's log for any abnormalities (log -r).

[*1] You can quickly check whether a switch supports Auto MDI-X: disconnect everything(!) and then plug a standard patch cable between any two ports. If the LEDs light up, Auto MDI-X is used.